├── .gitignore ├── README.md ├── _config.yml ├── docs ├── _config.yml └── index.md ├── lab ├── boosting │ ├── README.md │ ├── code │ │ ├── boosting.py │ │ ├── datasets.py │ │ ├── main.py │ │ └── utils.py │ └── tex │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis-imagelab.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── boosting.pdf │ │ ├── boosting.tex │ │ ├── img │ │ └── boosting │ │ │ ├── 1_boundary.pdf │ │ │ ├── 1_data.pdf │ │ │ ├── 2_boundary.pdf │ │ │ ├── 2_data.pdf │ │ │ ├── 3_boundary.pdf │ │ │ ├── 3_data.pdf │ │ │ └── gif │ │ │ ├── 000.pdf │ │ │ ├── 001.pdf │ │ │ ├── 002.pdf │ │ │ ├── 003.pdf │ │ │ ├── 004.pdf │ │ │ ├── 005.pdf │ │ │ ├── 006.pdf │ │ │ ├── 007.pdf │ │ │ ├── 008.pdf │ │ │ ├── 009.pdf │ │ │ ├── 010.pdf │ │ │ ├── 011.pdf │ │ │ ├── 012.pdf │ │ │ ├── 013.pdf │ │ │ ├── 014.pdf │ │ │ ├── 015.pdf │ │ │ ├── 016.pdf │ │ │ ├── 017.pdf │ │ │ ├── 018.pdf │ │ │ ├── 019.pdf │ │ │ ├── 020.pdf │ │ │ ├── 021.pdf │ │ │ ├── 022.pdf │ │ │ ├── 023.pdf │ │ │ ├── 024.pdf │ │ │ ├── 025.pdf │ │ │ ├── 026.pdf │ │ │ ├── 027.pdf │ │ │ ├── 028.pdf │ │ │ ├── 029.pdf │ │ │ ├── 030.pdf │ │ │ ├── 031.pdf │ │ │ ├── 032.pdf │ │ │ ├── 033.pdf │ │ │ ├── 034.pdf │ │ │ ├── 035.pdf │ │ │ ├── 036.pdf │ │ │ ├── 037.pdf │ │ │ ├── 038.pdf │ │ │ ├── 039.pdf │ │ │ ├── 040.pdf │ │ │ ├── 041.pdf │ │ │ ├── 042.pdf │ │ │ ├── 043.pdf │ │ │ ├── 044.pdf │ │ │ ├── 045.pdf │ │ │ ├── 046.pdf │ │ │ ├── 047.pdf │ │ │ ├── 048.pdf │ │ │ ├── 049.pdf │ │ │ ├── 050.pdf │ │ │ ├── 051.pdf │ │ │ ├── 052.pdf │ │ │ ├── 053.pdf │ │ │ ├── 054.pdf │ │ │ ├── 055.pdf │ │ │ ├── 056.pdf │ │ │ ├── 057.pdf │ │ │ ├── 058.pdf │ │ │ ├── 059.pdf │ │ │ ├── 060.pdf │ │ │ ├── 061.pdf │ │ │ ├── 062.pdf │ │ │ ├── 063.pdf │ │ │ ├── 064.pdf │ │ │ ├── 065.pdf │ │ │ ├── 066.pdf │ │ │ ├── 067.pdf │ │ │ ├── 068.pdf │ │ │ ├── 069.pdf │ │ │ ├── 070.pdf │ │ │ ├── 071.pdf │ │ │ ├── 072.pdf │ │ │ ├── 073.pdf │ │ │ ├── 074.pdf │ │ │ ├── 075.pdf │ │ │ ├── 076.pdf │ │ │ ├── 077.pdf │ │ │ ├── 078.pdf │ │ │ ├── 079.pdf │ │ │ ├── 080.pdf │ │ │ ├── 081.pdf │ │ │ ├── 082.pdf │ │ │ ├── 083.pdf │ │ │ ├── 084.pdf │ │ │ ├── 085.pdf │ │ │ ├── 086.pdf │ │ │ ├── 087.pdf │ │ │ ├── 088.pdf │ │ │ ├── 089.pdf │ │ │ ├── 090.pdf │ │ │ ├── 091.pdf │ │ │ ├── 092.pdf │ │ │ ├── 093.pdf │ │ │ ├── 094.pdf │ │ │ ├── 095.pdf │ │ │ ├── 096.pdf │ │ │ ├── 097.pdf │ │ │ ├── 098.pdf │ │ │ └── 099.pdf │ │ ├── logo_unimore_white.png │ │ ├── pgfplotsthemetol.sty │ │ └── src │ │ ├── algorithm.tex │ │ ├── boosting.tex │ │ ├── boundary1.tex │ │ ├── boundary2.tex │ │ ├── boundary3.tex │ │ ├── closer1.tex │ │ ├── closer2.tex │ │ ├── closer3.tex │ │ ├── decision_rule.tex │ │ ├── learning.tex │ │ ├── rationale.tex │ │ ├── setting.tex │ │ └── titlepage.tex ├── clustering │ ├── README.md │ ├── code │ │ ├── datasets.py │ │ ├── kmeans_clustering.py │ │ └── spectral_clustering.py │ └── tex │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis-imagelab.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── clustering.dvi │ │ ├── clustering.pdf │ │ ├── clustering.tex │ │ ├── clustering43.pdf │ │ ├── img │ │ ├── kmeans │ │ │ ├── emma.pdf │ │ │ ├── emma_segm.pdf │ │ │ ├── it01.pdf │ │ │ ├── it02.pdf │ │ │ ├── it03.pdf │ │ │ ├── it04.pdf │ │ │ ├── it05.pdf │ │ │ ├── it06.pdf │ │ │ ├── it07.pdf │ │ │ ├── it08.pdf │ │ │ ├── tm_fail.pdf │ │ │ └── tm_gt.pdf │ │ └── spectral │ │ │ ├── gt.pdf │ │ │ ├── pred.pdf │ │ │ ├── sigma_0.pdf │ │ │ ├── sigma_1.pdf │ │ │ ├── sigma_2.pdf │ │ │ ├── sigma_3.pdf │ │ │ └── sigma_4.pdf │ │ ├── logo_unimore_white.png │ │ ├── pgfplotsthemetol.sty │ │ └── src │ │ ├── kmeans_algo.tex │ │ ├── kmeans_drawbacks.tex │ │ ├── kmeans_intro.tex │ │ ├── kmeans_representation.tex │ │ ├── kmeans_segmentation - Copia.tex │ │ ├── kmeans_segmentation.tex │ │ ├── outline.tex │ │ ├── spectral_algo.tex │ │ ├── spectral_algo_1.tex │ │ ├── spectral_algo_2.tex │ │ ├── spectral_result.tex │ │ ├── spectral_sigma.tex │ │ ├── titlepage.tex │ │ └── useful_functions.tex ├── dimensionality_reduction │ ├── README.md │ ├── code │ │ ├── data_io.py │ │ ├── eigenfaces.py │ │ └── utils.py │ └── tex │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis-imagelab.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── dimensionality_reduction.pdf │ │ ├── dimensionality_reduction.tex │ │ ├── dimensionality_reduction43.pdf │ │ ├── img │ │ ├── eigenfaces │ │ │ ├── 0.pdf │ │ │ ├── 1.pdf │ │ │ ├── 2.pdf │ │ │ ├── 3.pdf │ │ │ ├── 4.pdf │ │ │ ├── accuracy.pdf │ │ │ ├── face_space.pdf │ │ │ ├── mean.pdf │ │ │ └── olivetti.pdf │ │ └── pca │ │ │ ├── components.pdf │ │ │ ├── original.pdf │ │ │ ├── original1.pdf │ │ │ ├── proj.pdf │ │ │ ├── proj1.pdf │ │ │ ├── reconstr.pdf │ │ │ └── reconstr1.pdf │ │ ├── logo_unimore_white.png │ │ ├── pgfplotsthemetol.sty │ │ └── src │ │ ├── accuracy.tex │ │ ├── eigenfaces_plots.tex │ │ ├── eigenfaces_predict.tex │ │ ├── eigenfaces_train.tex │ │ ├── eigenfaces_trick.tex │ │ ├── face_space.tex │ │ ├── outline.tex │ │ ├── pca_algo.tex │ │ ├── pca_algorithm.tex │ │ ├── pca_figure.tex │ │ ├── pca_intro.tex │ │ ├── pca_projections_1d.tex │ │ ├── pca_projections_2d.tex │ │ └── titlepage.tex ├── logistic_regression │ ├── README.md │ ├── code │ │ ├── data_io.py │ │ ├── logistic_regression.py │ │ ├── main.py │ │ ├── main_tf.py │ │ ├── merge_data.py │ │ └── visualization.py │ └── tex │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis-imagelab.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── img │ │ ├── books.jpg │ │ └── sgd │ │ │ ├── convex_function.png │ │ │ ├── descent │ │ │ ├── descent-0.png │ │ │ ├── descent-1.png │ │ │ ├── descent-2.png │ │ │ ├── descent-3.png │ │ │ ├── descent-4.png │ │ │ ├── descent-5.png │ │ │ ├── descent-6.png │ │ │ └── descent-7.png │ │ │ ├── level_sets.png │ │ │ ├── lr_too_big.jpg │ │ │ ├── lr_too_big.png │ │ │ ├── lr_too_small.jpg │ │ │ ├── lr_too_small.png │ │ │ └── non_convex_function.jpg │ │ ├── logistic_regression.dvi │ │ ├── logistic_regression.pdf │ │ ├── logistic_regression.tex │ │ ├── logo_unimore_white.png │ │ ├── pgfplotsthemetol.sty │ │ └── src │ │ ├── algorithm.tex │ │ ├── binary_crossentropy.tex │ │ ├── derive_loss_function.tex │ │ ├── game_of_thrones.tex │ │ ├── gradient.tex │ │ ├── gradient_descent_plz.tex │ │ ├── logistic_regression_prediction.tex │ │ ├── logistic_regression_rationale.tex │ │ ├── math_setting.tex │ │ ├── simplify_loss_function.tex │ │ └── titlepage.tex ├── naive_bayes │ ├── README.md │ ├── code │ │ ├── data_io.py │ │ ├── main.py │ │ └── naive_bayes.py │ └── tex │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis-imagelab.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── img │ │ ├── likelyhood.pdf │ │ ├── lkl_model.pdf │ │ ├── mnist.pdf │ │ └── test_image.pdf │ │ ├── logo_unimore_white.png │ │ ├── naive_bayes.dvi │ │ ├── naive_bayes.pdf │ │ ├── naive_bayes.tex │ │ ├── pgfplotsthemetol.sty │ │ └── src │ │ ├── algorithms.tex │ │ ├── bayes_classification.tex │ │ ├── image_prob.tex │ │ ├── inference.tex │ │ ├── likelyhood_test.tex │ │ ├── likelyhood_train.tex │ │ ├── logarithms.tex │ │ ├── mnist.tex │ │ ├── prior.tex │ │ ├── supervised.tex │ │ └── titlepage.tex ├── q_learning │ ├── README.md │ ├── code │ │ ├── action.py │ │ ├── agent.py │ │ ├── environment.py │ │ └── main.py │ └── tex │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis-imagelab.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── img │ │ ├── env.png │ │ └── gym.png │ │ ├── logo_unimore_white.png │ │ ├── pgfplotsthemetol.sty │ │ ├── q_learning.pdf │ │ ├── q_learning.tex │ │ └── src │ │ ├── empty_slide.tex │ │ ├── gym.tex │ │ ├── problem_setting.tex │ │ ├── q_learning_algorithm.tex │ │ ├── sarsa_algorithm.tex │ │ └── titlepage.tex ├── q_learning_function_approx │ ├── README.md │ └── code │ │ ├── circuit.m │ │ ├── circuit.py │ │ └── img │ │ ├── car.png │ │ ├── circuit_test.png │ │ ├── circuit_train.png │ │ ├── individualImage (1).png │ │ ├── individualImage (2).png │ │ ├── individualImage (3).png │ │ └── individualImage.png ├── support_vector_machines │ ├── README.md │ ├── code │ │ ├── datasets.py │ │ ├── main_pegasos.py │ │ ├── pegasos │ │ │ ├── SVM.py │ │ │ ├── __init__.py │ │ │ └── main.py │ │ ├── svm.py │ │ ├── svm_pegasos.py │ │ └── utils.py │ └── tex │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis-imagelab.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── img │ │ ├── people_classification │ │ │ ├── non_people.pdf │ │ │ └── people.pdf │ │ └── svm │ │ │ ├── linear_kernel_11.pdf │ │ │ ├── linear_kernel_12.pdf │ │ │ ├── linear_kernel_21.pdf │ │ │ ├── linear_kernel_22.pdf │ │ │ ├── max_margin.pdf │ │ │ ├── non_linear_data.pdf │ │ │ ├── rbf_kernel_11.pdf │ │ │ ├── rbf_kernel_12.pdf │ │ │ ├── rbf_kernel_21.pdf │ │ │ └── rbf_kernel_22.pdf │ │ ├── logo_unimore_white.png │ │ ├── pgfplotsthemetol.sty │ │ ├── src │ │ ├── agenda.tex │ │ ├── pegasos_algorithm.tex │ │ ├── pegasos_formulation.tex │ │ ├── pegasos_gradient.tex │ │ ├── people_classification_pipeline.tex │ │ ├── people_classification_setting.tex │ │ ├── subgradient.tex │ │ ├── svm_example_kernel_1.tex │ │ ├── svm_example_kernel_2.tex │ │ ├── svm_example_linear_1.tex │ │ ├── svm_example_linear_2.tex │ │ ├── svm_hard_margin.tex │ │ ├── svm_intro.tex │ │ ├── svm_kernels.tex │ │ ├── svm_kernels2.tex │ │ ├── svm_non_linear.tex │ │ ├── svm_setting.tex │ │ ├── svm_soft_margin.tex │ │ └── titlepage.tex │ │ ├── support_vector_machines.dvi │ │ ├── support_vector_machines.pdf │ │ └── support_vector_machines.tex ├── tensorflow_convolutional_nets │ ├── README.md │ ├── code │ │ ├── convnet_classification.py │ │ ├── convnet_segmentation.py │ │ └── utils.py │ └── tex │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── bibliography.bib │ │ ├── common │ │ ├── outline.tex │ │ └── titlepage.tex │ │ ├── img │ │ ├── template │ │ │ └── logo_unimore_white.png │ │ └── tf │ │ │ ├── mnist.png │ │ │ ├── tensorboard.jpg │ │ │ ├── tensorboard_tiles.png │ │ │ └── tiles_dataset.png │ │ ├── tensorflow_convnets.pdf │ │ └── tensorflow_convnets.tex ├── tensorflow_deep_autoencoders │ ├── README.md │ ├── code │ │ ├── dae_mnist.ipynb │ │ └── dae_mnist_sol.ipynb │ └── tex │ │ ├── Deep_Autoencoders.pdf │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── bibliography.bib │ │ ├── common │ │ ├── outline.tex │ │ └── titlepage.tex │ │ ├── img │ │ ├── autoencoder │ │ │ ├── accuracies.png │ │ │ ├── denoising.pdf │ │ │ ├── mnist_noisy.png │ │ │ ├── naive.pdf │ │ │ ├── reconstructions.png │ │ │ ├── semi.pdf │ │ │ └── semi.png │ │ └── template │ │ │ └── logo_unimore_white.png │ │ └── tensorflow_dae.tex ├── tensorflow_introduction │ ├── README.md │ └── tex │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── bibliography.bib │ │ ├── common │ │ ├── outline.tex │ │ └── titlepage.tex │ │ ├── img │ │ ├── template │ │ │ └── logo_unimore_white.png │ │ └── tf │ │ │ ├── computational_graph.png │ │ │ ├── graph_definition_0.jpg │ │ │ ├── graph_definition_1.jpg │ │ │ ├── graph_evaluation.jpg │ │ │ ├── placeholders.png │ │ │ ├── restore_variables.png │ │ │ ├── save_variables.png │ │ │ ├── tensorboard.jpg │ │ │ ├── tf_logo.png │ │ │ ├── variable_creation.jpg │ │ │ └── variable_initialization.png │ │ ├── tensorflow_intro.pdf │ │ └── tensorflow_intro.tex ├── tensorflow_linear_regression │ ├── README.md │ ├── code │ │ ├── linear_regression.py │ │ └── utils.py │ └── tex │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── bibliography.bib │ │ ├── common │ │ ├── outline.tex │ │ └── titlepage.tex │ │ ├── img │ │ └── template │ │ │ └── logo_unimore_white.png │ │ ├── tensorflow_regression.pdf │ │ └── tensorflow_regression.tex ├── tensorflow_mlp_numerical_data │ ├── data.py │ └── main.py ├── tensorflow_neural_network │ ├── README.md │ ├── code │ │ ├── neural_network.py │ │ └── utils.py │ └── tex │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── bibliography.bib │ │ ├── common │ │ ├── outline.tex │ │ └── titlepage.tex │ │ ├── img │ │ ├── template │ │ │ └── logo_unimore_white.png │ │ └── tf │ │ │ └── mnist.png │ │ ├── tensorflow_neural_nets.pdf │ │ └── tensorflow_neural_nets.tex ├── tensorflow_recurrent_nets │ ├── README.md │ ├── code │ │ ├── lstm_sequence_counting.py │ │ └── synthetic_dataset.py │ └── tex │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── bibliography.bib │ │ ├── common │ │ ├── outline.tex │ │ └── titlepage.tex │ │ ├── img │ │ └── template │ │ │ └── logo_unimore_white.png │ │ ├── tensorflow_lstm.pdf │ │ └── tensorflow_lstm.tex └── tensorflow_tweets │ └── code │ ├── lstm_tweet_classification.py │ └── utils.py ├── make_thumbs.py ├── slides ├── deep_learning │ ├── convolutional_neural_networks │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── bibliography.bib │ │ ├── common │ │ │ ├── credits.tex │ │ │ ├── outline.tex │ │ │ └── titlepage.tex │ │ ├── convolutional_neural_networks.pdf │ │ ├── convolutional_neural_networks.tex │ │ └── img │ │ │ ├── cnn │ │ │ ├── act_relu.jpg │ │ │ ├── act_sigmoid.jpg │ │ │ ├── act_tanh.jpg │ │ │ ├── activation_maps.jpg │ │ │ ├── black_box.png │ │ │ ├── busted_0.pdf │ │ │ ├── busted_1.pdf │ │ │ ├── cnn_architecture.jpg │ │ │ ├── cnn_vs_dnn.jpg │ │ │ ├── conv_animation │ │ │ │ ├── conv_animation-0.png │ │ │ │ ├── conv_animation-1.png │ │ │ │ ├── conv_animation-10.png │ │ │ │ ├── conv_animation-11.png │ │ │ │ ├── conv_animation-12.png │ │ │ │ ├── conv_animation-13.png │ │ │ │ ├── conv_animation-14.png │ │ │ │ ├── conv_animation-15.png │ │ │ │ ├── conv_animation-16.png │ │ │ │ ├── conv_animation-17.png │ │ │ │ ├── conv_animation-18.png │ │ │ │ ├── conv_animation-19.png │ │ │ │ ├── conv_animation-2.png │ │ │ │ ├── conv_animation-20.png │ │ │ │ ├── conv_animation-21.png │ │ │ │ ├── conv_animation-22.png │ │ │ │ ├── conv_animation-23.png │ │ │ │ ├── conv_animation-24.png │ │ │ │ ├── conv_animation-3.png │ │ │ │ ├── conv_animation-4.png │ │ │ │ ├── conv_animation-5.png │ │ │ │ ├── conv_animation-6.png │ │ │ │ ├── conv_animation-7.png │ │ │ │ ├── conv_animation-8.png │ │ │ │ └── conv_animation-9.png │ │ │ ├── conv_animation_stride │ │ │ │ ├── conv_animation_stride-0.png │ │ │ │ ├── conv_animation_stride-1.png │ │ │ │ ├── conv_animation_stride-2.png │ │ │ │ └── conv_animation_stride-3.png │ │ │ ├── convolution_gif.gif │ │ │ ├── convolution_gif_stride │ │ │ ├── convolution_gif_stride_2.gif │ │ │ ├── finetuning_rules.png │ │ │ ├── learned_weights.jpg │ │ │ ├── local_connectivity.jpg │ │ │ ├── maxpool.jpg │ │ │ ├── occlusions.jpg │ │ │ ├── pool.jpg │ │ │ ├── semseg_in.jpg │ │ │ ├── semseg_out.png │ │ │ ├── training_possibilities.png │ │ │ ├── tsne.jpg │ │ │ ├── vgg16_architecture.png │ │ │ ├── visualize_activation.png │ │ │ └── weights_inspection.png │ │ │ └── template │ │ │ └── logo_unimore_white.png │ ├── deep_neural_networks │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── bibliography.bib │ │ ├── common │ │ │ ├── credits.tex │ │ │ ├── outline.tex │ │ │ └── titlepage.tex │ │ ├── deep_neural_networks.pdf │ │ ├── deep_neural_networks.tex │ │ └── img │ │ │ ├── dnn │ │ │ ├── 2003_large_dataset.png │ │ │ ├── 2017_large_dataset.png │ │ │ ├── act_relu.jpg │ │ │ ├── act_sigmoid.jpg │ │ │ ├── act_tanh.jpg │ │ │ ├── augmentation.png │ │ │ ├── backprop │ │ │ │ ├── backprop.gif │ │ │ │ ├── backprop_0.png │ │ │ │ ├── backprop_1.png │ │ │ │ ├── backprop_2.png │ │ │ │ ├── backprop_3.png │ │ │ │ ├── backprop_4.png │ │ │ │ ├── backprop_5.png │ │ │ │ ├── backprop_6.png │ │ │ │ ├── backprop_7.png │ │ │ │ └── backprop_8.png │ │ │ ├── biological_neuron.jpg │ │ │ ├── cifar10.png │ │ │ ├── deep_learning_media.jpg │ │ │ ├── deep_learning_scale.png │ │ │ ├── dnn_overfit.jpg │ │ │ ├── dnn_overfit_reg.jpg │ │ │ ├── dropout_no.jpg │ │ │ ├── dropout_yes.jpg │ │ │ ├── forward_prop.jpg │ │ │ ├── linear_classifier.jpg │ │ │ ├── log_plot.png │ │ │ ├── neural_network.jpg │ │ │ ├── overfitting.png │ │ │ ├── single_neuron.jpg │ │ │ ├── single_neuron_backward.png │ │ │ └── single_neuron_forward.png │ │ │ └── template │ │ │ └── logo_unimore_white.png │ ├── gradient_descent │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── bibliography.bib │ │ ├── common │ │ │ ├── credits.tex │ │ │ ├── outline.tex │ │ │ └── titlepage.tex │ │ ├── gradient_descent.pdf │ │ ├── gradient_descent.tex │ │ └── img │ │ │ ├── sgd │ │ │ ├── convex_function.png │ │ │ ├── descent │ │ │ │ ├── descent-0.png │ │ │ │ ├── descent-1.png │ │ │ │ ├── descent-2.png │ │ │ │ ├── descent-3.png │ │ │ │ ├── descent-4.png │ │ │ │ ├── descent-5.png │ │ │ │ ├── descent-6.png │ │ │ │ └── descent-7.png │ │ │ ├── level_sets.png │ │ │ ├── lr_too_big.jpg │ │ │ ├── lr_too_big.png │ │ │ ├── lr_too_small.jpg │ │ │ ├── lr_too_small.png │ │ │ └── non_convex_function.jpg │ │ │ └── template │ │ │ └── logo_unimore_white.png │ └── recurrent_neural_networks │ │ ├── beamercolorthememetropolis-highcontrast.sty │ │ ├── beamercolorthememetropolis.sty │ │ ├── beamerfontthememetropolis.sty │ │ ├── beamerinnerthememetropolis.sty │ │ ├── beamerouterthememetropolis.sty │ │ ├── beamerthememetropolis.sty │ │ ├── bibliography.bib │ │ ├── common │ │ ├── credits.tex │ │ ├── outline.tex │ │ └── titlepage.tex │ │ ├── img │ │ ├── rnn │ │ │ ├── bptt.png │ │ │ ├── bptt_unroll.png │ │ │ ├── lstm_cell.png │ │ │ ├── rnn_api.png │ │ │ ├── rnn_unfold.png │ │ │ ├── vanilla_rnn.png │ │ │ └── vanilla_rnn_no_output.png │ │ └── template │ │ │ └── logo_unimore_white.png │ │ ├── recurrent_neural_networks.pdf │ │ └── recurrent_neural_networks.tex └── reinforcement_learning │ ├── 01_introduction_and_model_free_learning │ ├── 01_introduction_and_model_free_learning.pdf │ ├── 01_introduction_and_model_free_learning.tex │ ├── beamercolorthememetropolis-highcontrast.sty │ ├── beamercolorthememetropolis-imagelab.sty │ ├── beamercolorthememetropolis.sty │ ├── beamerfontthememetropolis.sty │ ├── beamerinnerthememetropolis.sty │ ├── beamerouterthememetropolis.sty │ ├── beamerthememetropolis.sty │ ├── img │ │ ├── agent_and_environment.pdf │ │ ├── blackjack.jpg │ │ ├── blackjack2.jpg │ │ ├── blackjack_mc_control.pdf │ │ ├── blackjack_value.pdf │ │ ├── doors.jpg │ │ ├── go_home_example.pdf │ │ ├── maze_model.pdf │ │ ├── maze_policy.pdf │ │ ├── maze_setup.pdf │ │ ├── maze_value.pdf │ │ ├── mc_control.pdf │ │ ├── mc_policy_iteration.pdf │ │ ├── ml_taxonomy.pdf │ │ ├── rl_taxonomy.pdf │ │ ├── sarsa.pdf │ │ ├── sarsa_control.pdf │ │ ├── sarsa_windy_gridworld.pdf │ │ └── windy_gridworld.pdf │ ├── logo_unimore_white.png │ └── src │ │ ├── acknowledge.tex │ │ ├── agent_and_environment.tex │ │ ├── agent_components.tex │ │ ├── bellman_equation.tex │ │ ├── bellman_equation_2.tex │ │ ├── blackjack_mc_control.tex │ │ ├── blackjack_reprise.tex │ │ ├── characteristics.tex │ │ ├── doors_example.tex │ │ ├── empty_slide.tex │ │ ├── epsilon_greedy.tex │ │ ├── go_home_example.tex │ │ ├── maze_environment.tex │ │ ├── maze_model.tex │ │ ├── maze_policy.tex │ │ ├── maze_value_function.tex │ │ ├── mc_algo.tex │ │ ├── mc_algo_incremental.tex │ │ ├── mc_blackjack_example.tex │ │ ├── mc_blackjack_value.tex │ │ ├── mc_control.tex │ │ ├── mc_intro.tex │ │ ├── mc_policy_evaluation.tex │ │ ├── mc_policy_iteration.tex │ │ ├── mc_vs_td_control.tex │ │ ├── ml_taxonomy.tex │ │ ├── model.tex │ │ ├── model_free_control.tex │ │ ├── model_free_prediction.tex │ │ ├── off_policy_learning.tex │ │ ├── on_off_policy.tex │ │ ├── policy.tex │ │ ├── policy_improv.tex │ │ ├── q_learning.tex │ │ ├── q_learning_2.tex │ │ ├── q_learning_algorithm.tex │ │ ├── return.tex │ │ ├── rewards.tex │ │ ├── rewards_examples.tex │ │ ├── rl_taxonomy.tex │ │ ├── sarsa.tex │ │ ├── sarsa_algorithm.tex │ │ ├── sarsa_control.tex │ │ ├── sarsa_windy_gridworld.tex │ │ ├── sequential_decision_making.tex │ │ ├── state.tex │ │ ├── states_and_observability.tex │ │ ├── td_learning_intro.tex │ │ ├── td_vs_mc.tex │ │ ├── td_vs_mc_2.tex │ │ ├── td_vs_mc_3.tex │ │ ├── titlepage.tex │ │ ├── value_function.tex │ │ └── windy_gridworld.tex │ ├── 02_function_approximation │ ├── 02_function_approximation.pdf │ ├── 02_function_approximation.tex │ ├── beamercolorthememetropolis-highcontrast.sty │ ├── beamercolorthememetropolis-imagelab.sty │ ├── beamercolorthememetropolis.sty │ ├── beamerfontthememetropolis.sty │ ├── beamerinnerthememetropolis.sty │ ├── beamerouterthememetropolis.sty │ ├── beamerthememetropolis.sty │ ├── img │ │ ├── dqn.JPG │ │ ├── dqn.pdf │ │ ├── dqn_results.pdf │ │ ├── gd.pdf │ │ ├── mountain_car.pdf │ │ ├── vfa_control.pdf │ │ └── vfa_types.pdf │ ├── logo_unimore_white.png │ └── src │ │ ├── acknowledge.tex │ │ ├── batch_rl.tex │ │ ├── convergence_control.tex │ │ ├── convergence_prediction.tex │ │ ├── dqn.tex │ │ ├── dqn_atari.tex │ │ ├── dqn_results.tex │ │ ├── empty_slide.tex │ │ ├── experience_replay.tex │ │ ├── feature_vectors.tex │ │ ├── gradient_descent.tex │ │ ├── incremental_control.tex │ │ ├── incremental_prediction.tex │ │ ├── large_scale_RL.tex │ │ ├── least_squares_prediction.tex │ │ ├── linear_q_approximation.tex │ │ ├── linear_vfa.tex │ │ ├── mc_vfa.tex │ │ ├── mountain_car.tex │ │ ├── q_approximation.tex │ │ ├── table_lookup.tex │ │ ├── td_vfa.tex │ │ ├── titlepage.tex │ │ ├── value_function_approximation.tex │ │ ├── vfa_by_sgd.tex │ │ ├── vfa_control.tex │ │ ├── vfa_types.tex │ │ ├── which_function.tex │ │ └── which_function_2.tex │ └── README.md └── thumbs ├── 01_introduction_and_model_free_learning.gif ├── 02_function_approximation.gif ├── boosting.gif ├── clustering.gif ├── convolutional_neural_networks.gif ├── deep_autoencoders.gif ├── deep_neural_networks.gif ├── dimensionality_reduction.gif ├── gradient_descent.gif ├── logistic_regression.gif ├── naive_bayes.gif ├── q_learning.gif ├── recurrent_neural_networks.gif ├── support_vector_machines.gif ├── tensorflow_convnets.gif ├── tensorflow_intro.gif ├── tensorflow_lstm.gif ├── tensorflow_neural_nets.gif └── tensorflow_regression.gif /.gitignore: -------------------------------------------------------------------------------- 1 | # Binary files 2 | *.pickle 3 | *.zip 4 | *.tar 5 | *.gz 6 | 7 | checkpoints/ 8 | 9 | # Python stuff 10 | *.pyc 11 | __pycache__/ 12 | 13 | # Pycharm stuff 14 | .idea/ 15 | 16 | # Texmaker stuff 17 | *.log 18 | *.nav 19 | *.out 20 | *.snm 21 | *.synctex.gz 22 | *.toc 23 | *.aux 24 | *.bbl 25 | *.blg 26 | *.vrb 27 | src/*.aux 28 | 29 | # UCL 30 | slides/reinforcement_learning/UCL 31 | -------------------------------------------------------------------------------- /_config.yml: -------------------------------------------------------------------------------- 1 | theme: jekyll-theme-cayman -------------------------------------------------------------------------------- /docs/_config.yml: -------------------------------------------------------------------------------- 1 | theme: jekyll-theme-cayman 2 | markdown: GFM 3 | -------------------------------------------------------------------------------- /lab/boosting/README.md: -------------------------------------------------------------------------------- 1 | ### Lab lecture - Boosting 2 | In this lecture we implement an 3 | Adaboost classifier in a Numpy environment, and test it on 4 | few two dimensional datasets (e.g. two moons etc.) -------------------------------------------------------------------------------- /lab/boosting/tex/boosting.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/boosting.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/1_boundary.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/1_boundary.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/1_data.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/1_data.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/2_boundary.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/2_boundary.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/2_data.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/2_data.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/3_boundary.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/3_boundary.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/3_data.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/3_data.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/000.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/000.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/001.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/001.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/002.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/002.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/003.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/003.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/004.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/004.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/005.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/005.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/006.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/006.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/007.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/007.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/008.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/008.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/009.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/009.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/010.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/010.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/011.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/011.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/012.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/012.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/013.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/013.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/014.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/014.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/015.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/015.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/016.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/016.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/017.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/017.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/018.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/018.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/019.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/019.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/020.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/020.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/021.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/021.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/022.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/022.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/023.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/023.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/024.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/024.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/025.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/025.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/026.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/026.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/027.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/027.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/028.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/028.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/029.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/029.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/030.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/030.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/031.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/031.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/032.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/032.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/033.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/033.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/034.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/034.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/035.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/035.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/036.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/036.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/037.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/037.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/038.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/038.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/039.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/039.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/040.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/040.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/041.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/041.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/042.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/042.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/043.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/043.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/044.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/044.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/045.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/045.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/046.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/046.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/047.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/047.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/048.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/048.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/049.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/049.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/050.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/050.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/051.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/051.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/052.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/052.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/053.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/053.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/054.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/054.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/055.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/055.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/056.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/056.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/057.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/057.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/058.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/058.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/059.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/059.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/060.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/060.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/061.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/061.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/062.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/062.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/063.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/063.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/064.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/064.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/065.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/065.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/066.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/066.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/067.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/067.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/068.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/068.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/069.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/069.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/070.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/070.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/071.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/071.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/072.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/072.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/073.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/073.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/074.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/074.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/075.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/075.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/076.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/076.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/077.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/077.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/078.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/078.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/079.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/079.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/080.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/080.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/081.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/081.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/082.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/082.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/083.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/083.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/084.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/084.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/085.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/085.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/086.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/086.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/087.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/087.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/088.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/088.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/089.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/089.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/090.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/090.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/091.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/091.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/092.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/092.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/093.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/093.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/094.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/094.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/095.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/095.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/096.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/096.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/097.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/097.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/098.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/098.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/img/boosting/gif/099.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/img/boosting/gif/099.pdf -------------------------------------------------------------------------------- /lab/boosting/tex/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/boosting/tex/logo_unimore_white.png -------------------------------------------------------------------------------- /lab/boosting/tex/src/boosting.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Boosting} 2 | Provides a general framework featuring: 3 | \begin{itemize} 4 | \uncover<1-2>{ 5 | \item weak classifiers: slightly better than random chance 6 | } 7 | \uncover<2>{ 8 | \begin{itemize} 9 | \item \textcolor{mImagelabRed}{How?} 10 | \end{itemize} 11 | } 12 | \uncover<1-2>{ 13 | \item cascade: each classifier is aware of previous ones errors 14 | } 15 | \uncover<2>{ 16 | \begin{itemize} 17 | \item \textcolor{mImagelabRed}{How?} 18 | \end{itemize} 19 | } 20 | \uncover<1-2>{ 21 | \item the prediction rule is based on all classifiers' predictions 22 | } 23 | \uncover<2>{ 24 | \begin{itemize} 25 | \item \textcolor{mImagelabRed}{How?} 26 | \end{itemize} 27 | } 28 | \end{itemize} 29 | \end{frame} 30 | -------------------------------------------------------------------------------- /lab/boosting/tex/src/boundary1.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Adaboost: example (1)} 2 | \begin{figure} 3 | \begin{tabular}{cc} 4 | \includegraphics[width=0.45\textwidth]{img/boosting/1_data.pdf}& 5 | \includegraphics[width=0.45\textwidth]{img/boosting/1_boundary.pdf} 6 | \end{tabular} 7 | \end{figure} 8 | \end{frame} 9 | -------------------------------------------------------------------------------- /lab/boosting/tex/src/boundary2.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Adaboost: example (2)} 2 | \begin{figure} 3 | \begin{tabular}{cc} 4 | \includegraphics[width=0.45\textwidth]{img/boosting/2_data.pdf}& 5 | \includegraphics[width=0.45\textwidth]{img/boosting/2_boundary.pdf} 6 | \end{tabular} 7 | \end{figure} 8 | \end{frame} 9 | -------------------------------------------------------------------------------- /lab/boosting/tex/src/boundary3.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Adaboost: example (3)} 2 | \begin{figure} 3 | \begin{tabular}{cc} 4 | \includegraphics[width=0.45\textwidth]{img/boosting/3_data.pdf}& 5 | \includegraphics[width=0.45\textwidth]{img/boosting/3_boundary.pdf} 6 | \end{tabular} 7 | \end{figure} 8 | \end{frame} 9 | -------------------------------------------------------------------------------- /lab/boosting/tex/src/closer1.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Adaboost: a closer look (1)} 2 | To build a weak classifier for $D_j$, repeat until $\epsilon_j<0.5$ 3 | \begin{itemize} 4 | \item randomly choose a data dimension 5 | \item randomly choose a threshold 6 | \item set all points above the threshold (along the chosen axis) to $1$ and all points below to $-1$ (or viceversa) 7 | \item compute $\epsilon_j$ over $D_j$ 8 | \end{itemize} 9 | \end{frame} 10 | -------------------------------------------------------------------------------- /lab/boosting/tex/src/closer2.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Adaboost: a closer look (2)} 2 | To compute the error and the efficiency of a weak classifier $M_j$ 3 | \begin{equation*} 4 | \epsilon_j = \sum_{M_j(x_i)\neq y_i} w_i 5 | \end{equation*} 6 | \begin{equation*} 7 | \alpha_j = \frac{1}{2} \ln\bigg(\frac{1-\epsilon_j}{\epsilon_j}\bigg) 8 | \end{equation*} 9 | \end{frame} 10 | -------------------------------------------------------------------------------- /lab/boosting/tex/src/closer3.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Adaboost: a closer look (3)} 2 | To update $w_i$: 3 | \begin{equation*} 4 | w_i^{(j+1)}=\frac{w_i^{(j)}}{Z^{(j)}} 5 | \begin{cases} 6 | \exp^{-\alpha_j} & \text{if $M_j(x_i)=y_i$} \\ 7 | \exp^{\alpha_j} & \text{if $M_j(x_i)\neq y_i$} 8 | \end{cases} 9 | \end{equation*} 10 | $Z^{(j)}$ is a normalization factor that ensures $\sum_i w_i = 1$ 11 | \end{frame} 12 | -------------------------------------------------------------------------------- /lab/boosting/tex/src/decision_rule.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Adaboost: decision rule} 2 | At test phase, you can classify an unknown vector $\vec{u}$ 3 | \begin{equation*} 4 | \mathcal{H}(\vec{u})=\text{sign}\Big(\sum_{j=1}^K \alpha_j M_j(\vec{u})\Big) 5 | \end{equation*} 6 | \end{frame} 7 | -------------------------------------------------------------------------------- /lab/boosting/tex/src/learning.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Adaboost: learning} 2 | \begin{center} 3 | \animategraphics[loop,controls,width=0.5\textwidth]{4}{img/boosting/gif/}{000}{099} 4 | \end{center} 5 | \end{frame} 6 | -------------------------------------------------------------------------------- /lab/boosting/tex/src/rationale.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Rationale} 2 | Boosting exploits the wisdom of the crowd: 3 | \begin{itemize} 4 | \item building many weak classifiers is easier than devising one complex prediction rule 5 | \item can I build a strong classifier out of a crowd of weak classifiers? 6 | \end{itemize} 7 | \end{frame} 8 | -------------------------------------------------------------------------------- /lab/boosting/tex/src/setting.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Adaboost: setting} 2 | We are given 3 | \begin{itemize} 4 | \item a dataset $D$ of examples $\{\vec{x_i}\}_{i=1}^N$ $\{y_i\}_{i=1}^N$, where $\vec{x_i} \in \mathbb{R}^d$ and $y_i \in \{-1, 1\}\quad\forall i=1,\ldots,N$ 5 | \item a predefined number of weak learners $K$ 6 | \end{itemize} 7 | \end{frame} 8 | -------------------------------------------------------------------------------- /lab/boosting/tex/src/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /lab/clustering/README.md: -------------------------------------------------------------------------------- 1 | ### Lab lecture - Clustering 2 | In this lecture we implement both K-means and spectral clustering, 3 | and test their both on several synthetic datasets (e.g. mixture 4 | of gaussians, two moons etc.) and on image pixels for the 5 | purpose of segmentation. -------------------------------------------------------------------------------- /lab/clustering/tex/clustering.dvi: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/clustering.dvi -------------------------------------------------------------------------------- /lab/clustering/tex/clustering.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/clustering.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/clustering43.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/clustering43.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/kmeans/emma.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/kmeans/emma.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/kmeans/emma_segm.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/kmeans/emma_segm.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/kmeans/it01.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/kmeans/it01.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/kmeans/it02.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/kmeans/it02.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/kmeans/it03.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/kmeans/it03.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/kmeans/it04.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/kmeans/it04.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/kmeans/it05.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/kmeans/it05.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/kmeans/it06.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/kmeans/it06.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/kmeans/it07.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/kmeans/it07.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/kmeans/it08.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/kmeans/it08.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/kmeans/tm_fail.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/kmeans/tm_fail.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/kmeans/tm_gt.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/kmeans/tm_gt.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/spectral/gt.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/spectral/gt.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/spectral/pred.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/spectral/pred.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/spectral/sigma_0.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/spectral/sigma_0.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/spectral/sigma_1.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/spectral/sigma_1.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/spectral/sigma_2.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/spectral/sigma_2.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/spectral/sigma_3.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/spectral/sigma_3.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/img/spectral/sigma_4.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/img/spectral/sigma_4.pdf -------------------------------------------------------------------------------- /lab/clustering/tex/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/clustering/tex/logo_unimore_white.png -------------------------------------------------------------------------------- /lab/clustering/tex/src/kmeans_algo.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{K-means: algorithm} 2 | The problem is NP-hard. A simple heuristic algorithm can be employed to converge to a \emph{local} minimum: 3 | \begin{itemize} 4 | \item Initialize $k$ centers randomly 5 | \item Repeat until convergence: 6 | \begin{itemize} 7 | \item assign each example to the closest center (i.e. lower euclidean distance) 8 | \item re-estimate centers as the mean of their clusters 9 | \end{itemize} 10 | \end{itemize} 11 | Try to implement it from scratch! 12 | \end{frame} -------------------------------------------------------------------------------- /lab/clustering/tex/src/kmeans_drawbacks.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Kmeans: limitations} 2 | \begin{itemize} 3 | \item it can get stuck into bad local minima 4 | \begin{itemize} 5 | \item OPTIONAL: run the algorithm many times and choose the most recurrent solution 6 | \end{itemize} 7 | \item can only be employed in spaces where the mean operation is defined 8 | \item due to its cost function, it can only cope with compact ball-shaped clusters 9 | \begin{figure} 10 | \begin{tabular}{cc} 11 | \small{GT} & \small{Result}\\ 12 | \includegraphics[width=0.35\textwidth]{img/kmeans/tm_gt.pdf}& 13 | \includegraphics[width=0.35\textwidth]{img/kmeans/tm_fail.pdf} 14 | \end{tabular} 15 | \end{figure} 16 | \end{itemize} 17 | \end{frame} -------------------------------------------------------------------------------- /lab/clustering/tex/src/kmeans_intro.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{K-means} 2 | \begin{itemize} 3 | \item Is a partitional clustering model 4 | \begin{itemize} 5 | \item splits data $\{x_i\}_1^n$ into $k$ disjoint sets 6 | \item the number of sets $k$ has to be provided as input 7 | \end{itemize} 8 | \item solves the following optimization problem: 9 | \begin{align*} 10 | \argmin_{\{c_1, \ldots, c_k\}} = \sum_{j=1}^k\sum_{i=1}^n \mathbf{I}(i,j)||x_i - c_j ||^2 11 | \end{align*} 12 | \begin{align*} 13 | \mathbf{I}(i,j)= 14 | \begin{cases} 15 | 1, & \text{$x_i$ belongs to cluster $j$}\\ 16 | 0, & \text{otherwise} 17 | \end{cases} 18 | \end{align*} 19 | \end{itemize} 20 | \end{frame} -------------------------------------------------------------------------------- /lab/clustering/tex/src/kmeans_segmentation - Copia.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{K-means} 3 | 4 | \end{frame} 5 | \egroup -------------------------------------------------------------------------------- /lab/clustering/tex/src/kmeans_segmentation.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Application: color segmentation} 2 | K-means can be employed for image segmentation, simply by grouping pixels in the color space. You can also add coordinates to each pixel to obtain a smooth output. 3 | \begin{figure} 4 | \begin{tabular}{cc} 5 | \small{Image} & \small{Segmentation}\\ 6 | \includegraphics[width=0.35\textwidth]{img/kmeans/emma.pdf}& 7 | \includegraphics[width=0.35\textwidth]{img/kmeans/emma_segm.pdf} 8 | \end{tabular} 9 | \end{figure} 10 | Try it! 11 | \end{frame} -------------------------------------------------------------------------------- /lab/clustering/tex/src/outline.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Agenda} 2 | \tableofcontents 3 | \end{frame} 4 | -------------------------------------------------------------------------------- /lab/clustering/tex/src/spectral_algo.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Spectral clustering} 3 | 4 | \end{frame} 5 | \egroup -------------------------------------------------------------------------------- /lab/clustering/tex/src/spectral_algo_1.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Spectral clustering: algorithm (1)} 2 | Clustering model based on the spectral graph theory. 3 | \begin{itemize} 4 | \item build a graph over examples, representing it with the adjacency matrix $A$ 5 | \begin{equation*} 6 | A_{i,j} = e^{-\frac{\sum_{k=1}^d||x_i^k-x_j^k||^2}{\sigma^2}} 7 | \end{equation*} 8 | \item build the degree matrix $D$ of the graph. It is a diagonal matrix holding for each element the sum of the incoming adges. 9 | \item compute the normalized laplacian $L$ 10 | \begin{equation} 11 | L=I-D^{-0.5}AD^{-0.5} 12 | \end{equation} 13 | \end{itemize} 14 | \end{frame} -------------------------------------------------------------------------------- /lab/clustering/tex/src/spectral_algo_2.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Spectral clustering: algorithm (2)} 2 | \begin{itemize} 3 | \item Compute the eigenvectors and sort them for increasing eigenvalues 4 | \item Choose the eigenvectors \underline{from the second to the desired number of clusters} 5 | \item Those eigenvectors provide a representation of data in a fancy embedding space: run K-means over such eigenvectors. 6 | \end{itemize} 7 | \end{frame} -------------------------------------------------------------------------------- /lab/clustering/tex/src/spectral_result.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Spectral clustering: result} 3 | \setlength{\tabcolsep}{.07em} 4 | \begin{figure} 5 | \begin{tabular}{cc} 6 | \small{GT} & \small{Result}\\ 7 | \includegraphics[width=0.45\textwidth]{img/spectral/gt.pdf}& 8 | \includegraphics[width=0.45\textwidth]{img/spectral/pred.pdf} 9 | \end{tabular} 10 | \end{figure} 11 | \end{frame} 12 | \egroup -------------------------------------------------------------------------------- /lab/clustering/tex/src/spectral_sigma.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Beware: $\sigma$ counts (in large amounts!)} 3 | \setlength{\tabcolsep}{.07em} 4 | \begin{figure} 5 | \begin{tabular}{ccc} 6 | \small{$\sigma=0.01$} & \small{$\sigma=0.1$} & \small{$\sigma=1$}\\ 7 | \includegraphics[width=0.3\textwidth]{img/spectral/sigma_0.pdf}& 8 | \includegraphics[width=0.3\textwidth]{img/spectral/sigma_2.pdf}& 9 | \includegraphics[width=0.3\textwidth]{img/spectral/sigma_4.pdf} 10 | \end{tabular} 11 | \end{figure} 12 | \end{frame} 13 | \egroup -------------------------------------------------------------------------------- /lab/clustering/tex/src/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /lab/dimensionality_reduction/README.md: -------------------------------------------------------------------------------- 1 | ### Lab lecture - Dimensionality reduction 2 | In this lecture we implement PCA and test it on a 3D dataset. 3 | Moreover, we implement Eigenfaces on the Olivetti dataset for 4 | face recognition. -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/dimensionality_reduction.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/dimensionality_reduction.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/dimensionality_reduction43.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/dimensionality_reduction43.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/eigenfaces/0.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/eigenfaces/0.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/eigenfaces/1.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/eigenfaces/1.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/eigenfaces/2.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/eigenfaces/2.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/eigenfaces/3.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/eigenfaces/3.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/eigenfaces/4.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/eigenfaces/4.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/eigenfaces/accuracy.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/eigenfaces/accuracy.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/eigenfaces/face_space.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/eigenfaces/face_space.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/eigenfaces/mean.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/eigenfaces/mean.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/eigenfaces/olivetti.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/eigenfaces/olivetti.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/pca/components.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/pca/components.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/pca/original.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/pca/original.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/pca/original1.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/pca/original1.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/pca/proj.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/pca/proj.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/pca/proj1.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/pca/proj1.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/pca/reconstr.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/pca/reconstr.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/img/pca/reconstr1.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/img/pca/reconstr1.pdf -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/dimensionality_reduction/tex/logo_unimore_white.png -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/src/accuracy.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Eigenfaces: how many dimensions?} 2 | \centering 3 | \includegraphics[width=0.5\textwidth]{img/eigenfaces/accuracy.pdf} 4 | \end{frame} 5 | -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/src/eigenfaces_predict.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Eigenfaces} 2 | To classify a test image: 3 | \begin{itemize} 4 | \item Project the image in the reduced spaces built in the training phase 5 | \item Perform \textbf{nearest neighbor classification}: 6 | \begin{itemize} 7 | \item Roughly speaking, \underline{choose the class of the nearest training example} (in the reduced space) 8 | \end{itemize} 9 | \end{itemize} 10 | \end{frame} -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/src/eigenfaces_train.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Eigenfaces} 2 | Famous algorithm for face recognition. Training is as simple as: 3 | \begin{itemize} 4 | \item load faces and annotations from the Olivetti dataset (datasets.get\_faces\_dataset takes care of loading and flattening images)\\ 5 | \begin{center} 6 | \includegraphics[width=0.2\textwidth]{img/eigenfaces/olivetti.pdf} 7 | \end{center} 8 | \item Select a number of principal components and fit a PCA on training faces 9 | \end{itemize} 10 | \end{frame} -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/src/eigenfaces_trick.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Eigenfaces: a magic trick to compute eigenvectors} 2 | Each Olivetti image is $112\times 92$. Once flattened, is a vector of $10304$ pixels: 3 | \begin{itemize} 4 | \item The convariance matrix is $10304\times 10304$ 5 | \item Computing eigenvectors and eigenvalues \underline{is a pain} 6 | \item Instead, compute the covariance matrix of transposed $X$: 7 | \begin{equation*} 8 | \Sigma = (X-\mu)(X-\mu)^T 9 | \end{equation*} 10 | \item Once selected the principal components $\tilde{E}$ of this weirdo space, you can compute the original eigenvectors just like: 11 | \begin{equation*} 12 | E = X^T\cdot \tilde{E} 13 | \end{equation*} 14 | \end{itemize} 15 | \end{frame} -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/src/face_space.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Eigenfaces: face space} 2 | \centering 3 | \includegraphics[width=0.5\textwidth]{img/eigenfaces/face_space.pdf} 4 | \end{frame} 5 | -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/src/outline.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Agenda} 2 | \tableofcontents 3 | \end{frame} 4 | -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/src/pca_algo.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{PCA: projecting and reconstructing} 2 | \begin{figure} 3 | \begin{tabular}{ccc} 4 | \small{original data} & \small{projection} & \small{reconstruction}\\ 5 | \includegraphics[width=0.3\textwidth]{img/pca/original.pdf}& 6 | \includegraphics[width=0.3\textwidth]{img/pca/proj.pdf}& 7 | \includegraphics[width=0.3\textwidth]{img/pca/reconstr.pdf} 8 | \end{tabular} 9 | \end{figure} 10 | \end{frame} -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/src/pca_figure.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{PCA: plotting components} 2 | \centering 3 | \includegraphics[width=0.5\textwidth]{img/pca/components.pdf} 4 | \end{frame} 5 | -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/src/pca_intro.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{PCA} 2 | \begin{itemize} 3 | \item Linear dimensionality reduction model 4 | \begin{itemize} 5 | \item Subspace projection is linear 6 | \item Reconstruction is linear 7 | \end{itemize} 8 | \item Projects data in a new space subject to: 9 | \begin{itemize} 10 | \item the direction exhibiting highest variance in feature space is projected on the first axis, the one exhibiting the second highest variance on the second axis, and so on. 11 | \item axis of the new space are orthogonal (covariance is zero). 12 | \end{itemize} 13 | \end{itemize} 14 | \end{frame} 15 | -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/src/pca_projections_1d.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{PCA: projecting and reconstructing (1D)} 2 | \begin{figure} 3 | \begin{tabular}{ccc} 4 | \small{original data} & \small{projection} & \small{reconstruction}\\ 5 | \includegraphics[width=0.3\textwidth]{img/pca/original1.pdf}& 6 | \includegraphics[width=0.3\textwidth]{img/pca/proj1.pdf}& 7 | \includegraphics[width=0.3\textwidth]{img/pca/reconstr1.pdf} 8 | \end{tabular} 9 | \end{figure} 10 | \end{frame} -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/src/pca_projections_2d.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{PCA: projecting and reconstructing (2D)} 2 | \begin{figure} 3 | \begin{tabular}{ccc} 4 | \small{original data} & \small{projection} & \small{reconstruction}\\ 5 | \includegraphics[width=0.3\textwidth]{img/pca/original.pdf}& 6 | \includegraphics[width=0.3\textwidth]{img/pca/proj.pdf}& 7 | \includegraphics[width=0.3\textwidth]{img/pca/reconstr.pdf} 8 | \end{tabular} 9 | \end{figure} 10 | \end{frame} -------------------------------------------------------------------------------- /lab/dimensionality_reduction/tex/src/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /lab/logistic_regression/README.md: -------------------------------------------------------------------------------- 1 | ### Lab lecture - Logistic Regression 2 | In this lecture we implement a Logistic Regression classifier 3 | and we train it by gradient descent. The dataset we use is taken 4 | from Kaggle, and regards predicting deaths in "A Song of Ice and Fire" 5 | books. -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/books.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/books.jpg -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/convex_function.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/convex_function.png -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/descent/descent-0.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/descent/descent-0.png -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/descent/descent-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/descent/descent-1.png -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/descent/descent-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/descent/descent-2.png -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/descent/descent-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/descent/descent-3.png -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/descent/descent-4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/descent/descent-4.png -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/descent/descent-5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/descent/descent-5.png -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/descent/descent-6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/descent/descent-6.png -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/descent/descent-7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/descent/descent-7.png -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/level_sets.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/level_sets.png -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/lr_too_big.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/lr_too_big.jpg -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/lr_too_big.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/lr_too_big.png -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/lr_too_small.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/lr_too_small.jpg -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/lr_too_small.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/lr_too_small.png -------------------------------------------------------------------------------- /lab/logistic_regression/tex/img/sgd/non_convex_function.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/img/sgd/non_convex_function.jpg -------------------------------------------------------------------------------- /lab/logistic_regression/tex/logistic_regression.dvi: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/logistic_regression.dvi -------------------------------------------------------------------------------- /lab/logistic_regression/tex/logistic_regression.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/logistic_regression.pdf -------------------------------------------------------------------------------- /lab/logistic_regression/tex/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/logistic_regression/tex/logo_unimore_white.png -------------------------------------------------------------------------------- /lab/logistic_regression/tex/src/binary_crossentropy.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Binary crossentropy loss} 3 | During training, we want to \textbf{minimize} the following function: 4 | \begin{equation*} 5 | \mathcal{L}(w) = -\frac{1}{N} \sum_{i=1}^N [Y_i \log(F(X_i, w)) + (1 - Y_i)\log(1-F(X_i, w))] 6 | \end{equation*} 7 | \end{frame} 8 | \egroup -------------------------------------------------------------------------------- /lab/logistic_regression/tex/src/derive_loss_function.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Derive the loss function} 3 | Back to our problem. We need to take the derivative of this function w.r.t. $w$: 4 | \begin{equation*} 5 | \mathcal{L}(w) = -\frac{1}{N} \sum_{i=1}^N \bigg[Y_i(w^T\cdot X_i) - \log\bigg(1+e^{w^T\cdot X_i}\bigg)\bigg] 6 | \end{equation*} 7 | \begin{align*} 8 | \frac{\delta \mathcal{L}(w)}{\delta w_j} &= -\frac{1}{N} \sum_{i=1}^N \bigg[Y_ix_i^{(j)} - \frac{e^{w^T\cdot X_i}}{1 + e^{w^T\cdot X_i}}x_i^{(j)}\bigg]\\ 9 | &=- \frac{1}{N} \sum_{i=1}^N \bigg[Y_i - \frac{e^{w^T\cdot X_i}}{1 + e^{w^T\cdot X_i}}\bigg]x_i^{(j)}\\ 10 | &=- \frac{1}{N} \sum_{i=1}^N \bigg[Y_i - F(X_i, w)\bigg]x_i^{(j)} 11 | \end{align*} 12 | \end{frame} 13 | \egroup -------------------------------------------------------------------------------- /lab/logistic_regression/tex/src/game_of_thrones.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Today's case study} 3 | \begin{minipage}{0.45\textwidth} 4 | \includegraphics[width=\textwidth]{img/books.jpg} 5 | \end{minipage} 6 | \begin{minipage}{0.45\textwidth} 7 | \begin{itemize} 8 | \item We want to predict if a character is alive or dead; 9 | \item (Some) of our features are: 10 | \begin{itemize} 11 | \item male or female; 12 | \item married or not; 13 | \item number of deaths witnessed; 14 | \item number of dead relatives; 15 | \item ... and many more. 16 | \end{itemize} 17 | \end{itemize} 18 | \end{minipage} 19 | \end{frame} 20 | \egroup -------------------------------------------------------------------------------- /lab/logistic_regression/tex/src/logistic_regression_rationale.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Logistic regression} 3 | \only<1>{ 4 | We need to learn a function that maps X to Y such that ``it works well on the training set''. 5 | } 6 | 7 | \only<2>{ 8 | We need to learn the parameters $\textbf{w}$ of a parametric function that maps X to Y such that ``it works well on the training set''. 9 | } 10 | 11 | \only<3>{ 12 | We need to learn the parameters $\textbf{w}$ of a parametric function that maps X to Y such that \textbf{some error is as low as possible on the training set}. 13 | } 14 | 15 | 16 | \end{frame} 17 | \egroup -------------------------------------------------------------------------------- /lab/logistic_regression/tex/src/math_setting.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Supervised learning setting} 3 | We are given a training set $\{X_i, Y_i\}_{i=1}^N$, with $X_i \in \mathbb{R}^m$ and $Y_i \in \{0, 1\}$ for each $i=1,\ldots,N$. 4 | \begin{itemize} 5 | \item $N$ is the number of training examples; 6 | \item each example $X_i = \{x_i^{(1)}, \ldots , x_i^{(m)}\}$ is a vector of $m$ features; 7 | \item each label $Y_i$ is either 0 or 1. 8 | \end{itemize} 9 | \end{frame} 10 | \egroup -------------------------------------------------------------------------------- /lab/logistic_regression/tex/src/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /lab/naive_bayes/README.md: -------------------------------------------------------------------------------- 1 | ### Lab lecture - Naive Bayes 2 | In this lecture we implement a Naive Bayes Classifier for image 3 | classification. We employ two datasets for the purpose: MNIST and 4 | Fashion MNIST. -------------------------------------------------------------------------------- /lab/naive_bayes/tex/img/likelyhood.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/naive_bayes/tex/img/likelyhood.pdf -------------------------------------------------------------------------------- /lab/naive_bayes/tex/img/lkl_model.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/naive_bayes/tex/img/lkl_model.pdf -------------------------------------------------------------------------------- /lab/naive_bayes/tex/img/mnist.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/naive_bayes/tex/img/mnist.pdf -------------------------------------------------------------------------------- /lab/naive_bayes/tex/img/test_image.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/naive_bayes/tex/img/test_image.pdf -------------------------------------------------------------------------------- /lab/naive_bayes/tex/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/naive_bayes/tex/logo_unimore_white.png -------------------------------------------------------------------------------- /lab/naive_bayes/tex/naive_bayes.dvi: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/naive_bayes/tex/naive_bayes.dvi -------------------------------------------------------------------------------- /lab/naive_bayes/tex/naive_bayes.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/naive_bayes/tex/naive_bayes.pdf -------------------------------------------------------------------------------- /lab/naive_bayes/tex/src/image_prob.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Image probability $P(X)$} 3 | It is computed the same way as the likelyhood, but estimating the model without conditioning on a particolar class. 4 | 5 | Just use the whole dataset. 6 | \end{frame} 7 | \egroup -------------------------------------------------------------------------------- /lab/naive_bayes/tex/src/inference.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Inference} 3 | Once we can model $P(Y/X)$ for each class, we can classify an image by choosing the class that maximizes it: 4 | \begin{equation*} 5 | \tilde{Y} = \argmax_{Y} P(Y/X) = P(X/Y)P(Y) 6 | \end{equation*} 7 | \end{frame} 8 | \egroup -------------------------------------------------------------------------------- /lab/naive_bayes/tex/src/likelyhood_test.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Likelyhood $P(X/Y)$} 3 | During testing, the likelyhood of an image under a class is built by taking the class model (the one built before, during training) and: 4 | \begin{itemize} 5 | \item for each active pixel, account for the probability of being active given class; 6 | \item for each zero pixel, account for one minus the probability of being active given class; 7 | \end{itemize} 8 | \begin{minipage}{0.25\textwidth} 9 | \centering 10 | \includegraphics[width=\textwidth]{img/test_image.pdf} 11 | \end{minipage} 12 | \begin{minipage}{0.65\textwidth} 13 | \centering 14 | \includegraphics[width=\textwidth]{img/likelyhood.pdf} 15 | \end{minipage} 16 | \end{frame} 17 | \egroup -------------------------------------------------------------------------------- /lab/naive_bayes/tex/src/likelyhood_train.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Likelyhood $P(X/Y)$} 3 | \textbf{Naive assumption}: all pixels are independent given the class. The probability of an image is the product of the probability of every single pixel.\\ 4 | \begin{equation*} 5 | P(X_i/Y_c)=\prod_{j=1}^m P(x_i^{(j)}/Y_c) 6 | \end{equation*} 7 | During training, we need to model this for each possible class: 8 | \includegraphics[width=\textwidth]{img/lkl_model.pdf} 9 | \end{frame} 10 | \egroup -------------------------------------------------------------------------------- /lab/naive_bayes/tex/src/logarithms.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Use logarithms!} 3 | All those products of probabilities can result in nothing. Use log-probabilities instead! 4 | \begin{equation*} 5 | \log{P(X_i/Y_c)}=\sum_{j=1}^m \log{P(x_i^{(j)}/Y_c)} 6 | \end{equation*} 7 | \begin{equation*} 8 | \log{P(Y/X)} = \log{P(X/Y)} + \log{P(Y)} 9 | \end{equation*} 10 | \end{frame} 11 | \egroup -------------------------------------------------------------------------------- /lab/naive_bayes/tex/src/mnist.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{MNIST digits classification} 3 | Today we meet MNIST for the first time: 4 | 5 | \begin{minipage}{0.45\textwidth} 6 | \centering 7 | \includegraphics[width=\textwidth]{img/mnist.pdf} 8 | \end{minipage} 9 | \begin{minipage}{0.45\textwidth} 10 | \begin{itemize} 11 | \item 70000 images of handwritten images; 12 | \item available grayscale, we will use a binarized version; 13 | \item the task is to classify each image into the right digit. 14 | \end{itemize} 15 | \end{minipage} 16 | 17 | \end{frame} 18 | \egroup -------------------------------------------------------------------------------- /lab/naive_bayes/tex/src/prior.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Prior $P(Y)$} 3 | Finding the prior of a class $Y$ is easy.\\ 4 | Simply count the number of examples of each class and divide by the number of total examples. 5 | 6 | \begin{equation*} 7 | P(Y=c) = \frac{\sum_{i=1}^n \bold{1}\{Y_i == c\}}{n} 8 | \end{equation*} 9 | \end{frame} 10 | \egroup -------------------------------------------------------------------------------- /lab/naive_bayes/tex/src/supervised.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Supervised learning setting} 3 | We are given a training set $\{X_i, Y_i\}_{i=1}^n$, with $X_i \in \mathbb{R}^m$ and $Y_i \in \mathbb{R}$ for each $i=1,\ldots,n$. 4 | \begin{itemize} 5 | \item $n$ is the number of training images; 6 | \item each image $X_i = \{x_i^{(1)}, \ldots , x_i^{(m)}\}$ is a vector of $m$ pixels; 7 | \item each label $Y_i$ is just a number among $\{1, \ldots, d\}$. 8 | \end{itemize} 9 | \end{frame} 10 | \egroup -------------------------------------------------------------------------------- /lab/naive_bayes/tex/src/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /lab/q_learning/README.md: -------------------------------------------------------------------------------- 1 | ### Lab lecture - Q Learning 2 | In this lecture, we teach an agent to escape a maze by Reinforcement 3 | Learning. Both the environment and the agent are Numpy based. -------------------------------------------------------------------------------- /lab/q_learning/tex/img/env.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/q_learning/tex/img/env.png -------------------------------------------------------------------------------- /lab/q_learning/tex/img/gym.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/q_learning/tex/img/gym.png -------------------------------------------------------------------------------- /lab/q_learning/tex/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/q_learning/tex/logo_unimore_white.png -------------------------------------------------------------------------------- /lab/q_learning/tex/q_learning.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/q_learning/tex/q_learning.pdf -------------------------------------------------------------------------------- /lab/q_learning/tex/src/empty_slide.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{} 3 | \end{frame} 4 | \egroup -------------------------------------------------------------------------------- /lab/q_learning/tex/src/gym.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{More fun with gym!} 3 | If you are curious about RL, try \href{https://github.com/openai/gym}{gym}: 4 | \begin{center} 5 | pip install gym 6 | \end{center} 7 | \centering 8 | \includegraphics[width=0.5\textwidth]{img/gym.png} 9 | \end{frame} 10 | \egroup -------------------------------------------------------------------------------- /lab/q_learning/tex/src/problem_setting.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Problem setting} 3 | \begin{minipage}{0.5\textwidth} 4 | \includegraphics[width=0.9\textwidth]{img/env.png} 5 | \end{minipage} 6 | \begin{minipage}{0.45\textwidth} 7 | We have an agent stuck in a maze. 8 | \begin{itemize} 9 | \item state is (x,y) position 10 | \item reward is -1 for each time step 11 | \item when the exit of the labirinth is reached, the episode terminates 12 | \item allowed actions are N,S,W,E 13 | \end{itemize} 14 | Guide him out with reinforcement learning! 15 | \end{minipage} 16 | \end{frame} 17 | \egroup -------------------------------------------------------------------------------- /lab/q_learning/tex/src/q_learning_algorithm.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Q-learning algorithm for off-policy control} 3 | \begin{algorithmic} 4 | \STATE Initialize $Q(s,a), \forall s \in S, a \in \mathcal{A}(s)$, arbitrarily, and $Q(terminal-state, \dot)=0$ 5 | \FOR{each episode} 6 | \STATE Intialise $S$ 7 | \FOR {each step of episode} 8 | \STATE Choose $A$ from $S$ using policy derived from $Q$ (e.g., $\epsilon$-greedy) 9 | \STATE Take action $A$, observe $R$, $S^{\prime}$ 10 | \STATE $Q(S,A) \leftarrow Q(S,A) + \alpha (R + \gamma \max_{a^{\prime}}Q(S^{\prime}, a^{\prime}) - Q(S,A))$ 11 | \STATE $S \leftarrow S^{\prime}$ 12 | \ENDFOR 13 | \ENDFOR 14 | \end{algorithmic} 15 | \end{frame} 16 | \egroup -------------------------------------------------------------------------------- /lab/q_learning/tex/src/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /lab/q_learning_function_approx/README.md: -------------------------------------------------------------------------------- 1 | ### Lab lecture - Q-Learning with function approximation 2 | In this lecture we teach a car to drive around within a circuit. 3 | We use function approximation to generalize to unseen circuits. 4 | 5 | --- 6 | **Please note: this lecture is incomplete** -------------------------------------------------------------------------------- /lab/q_learning_function_approx/code/img/car.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/q_learning_function_approx/code/img/car.png -------------------------------------------------------------------------------- /lab/q_learning_function_approx/code/img/circuit_test.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/q_learning_function_approx/code/img/circuit_test.png -------------------------------------------------------------------------------- /lab/q_learning_function_approx/code/img/circuit_train.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/q_learning_function_approx/code/img/circuit_train.png -------------------------------------------------------------------------------- /lab/q_learning_function_approx/code/img/individualImage (1).png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/q_learning_function_approx/code/img/individualImage (1).png -------------------------------------------------------------------------------- /lab/q_learning_function_approx/code/img/individualImage (2).png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/q_learning_function_approx/code/img/individualImage (2).png -------------------------------------------------------------------------------- /lab/q_learning_function_approx/code/img/individualImage (3).png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/q_learning_function_approx/code/img/individualImage (3).png -------------------------------------------------------------------------------- /lab/q_learning_function_approx/code/img/individualImage.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/q_learning_function_approx/code/img/individualImage.png -------------------------------------------------------------------------------- /lab/support_vector_machines/README.md: -------------------------------------------------------------------------------- 1 | ### Lab lecture - Support Vector Machines 2 | In this lecture we use the Sklearn implementation of SVMs to perform 3 | people vs non-people classification, with HoG features. -------------------------------------------------------------------------------- /lab/support_vector_machines/code/pegasos/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/code/pegasos/__init__.py -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/img/people_classification/non_people.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/img/people_classification/non_people.pdf -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/img/people_classification/people.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/img/people_classification/people.pdf -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/img/svm/linear_kernel_11.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/img/svm/linear_kernel_11.pdf -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/img/svm/linear_kernel_12.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/img/svm/linear_kernel_12.pdf -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/img/svm/linear_kernel_21.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/img/svm/linear_kernel_21.pdf -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/img/svm/linear_kernel_22.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/img/svm/linear_kernel_22.pdf -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/img/svm/max_margin.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/img/svm/max_margin.pdf -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/img/svm/non_linear_data.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/img/svm/non_linear_data.pdf -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/img/svm/rbf_kernel_11.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/img/svm/rbf_kernel_11.pdf -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/img/svm/rbf_kernel_12.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/img/svm/rbf_kernel_12.pdf -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/img/svm/rbf_kernel_21.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/img/svm/rbf_kernel_21.pdf -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/img/svm/rbf_kernel_22.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/img/svm/rbf_kernel_22.pdf -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/logo_unimore_white.png -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/src/agenda.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Today's agenda} 2 | \tableofcontents 3 | \end{frame} -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/src/pegasos_algorithm.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{PEGASOS algorithm} 2 | \begin{algorithm}[H] 3 | \begin{algorithmic}[1] 4 | \STATE $\vec{w}^0 = \vec{0}$, $t=0$ 5 | \FOR{epoch in $1,\ldots,\text{n\_epochs}$} 6 | \FOR{$j=1,\ldots,N$} 7 | \STATE $t=t+1$ 8 | \STATE $\eta^t=\frac{1}{t\lambda}$ 9 | \IF{$y_j \langle \vec{w}_t \vec{x}_j \rangle < 1$} 10 | \STATE $\vec{w}^{t+1}=(1-\eta^t\lambda)\vec{w}^t+\eta^ty_jx_j$ 11 | \ELSE 12 | \STATE $\vec{w}^{t+1}=(1-\eta^t\lambda)\vec{w}^t$ 13 | \ENDIF 14 | \ENDFOR 15 | \ENDFOR 16 | \end{algorithmic} 17 | \caption{PEGASOS} 18 | \label{alg:pegasos_train} 19 | \end{algorithm} 20 | \end{frame} -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/src/people_classification_setting.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{SVM for people detection} 2 | We will discriminate between people and non people 3 | \begin{figure} 4 | \begin{tabular}{cc} 5 | \small{people} & \small{non people}\\ 6 | \includegraphics[width=0.4\textwidth]{img/people_classification/people.pdf}& 7 | \includegraphics[width=0.4\textwidth]{img/people_classification/non_people.pdf} 8 | \end{tabular} 9 | \end{figure} 10 | \end{frame} -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/src/svm_example_kernel_1.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{SVM example} 2 | Using a rbf (gaussian) kernel: 3 | \begin{figure} 4 | \begin{tabular}{cc} 5 | \includegraphics[width=0.4\textwidth]{img/svm/rbf_kernel_11.pdf}& 6 | \includegraphics[width=0.4\textwidth]{img/svm/rbf_kernel_12.pdf} 7 | \end{tabular} 8 | \end{figure} 9 | \end{frame} -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/src/svm_example_kernel_2.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{SVM example (2)} 2 | Using a rbf (gaussian) kernel: 3 | \begin{figure} 4 | \begin{tabular}{cc} 5 | \includegraphics[width=0.4\textwidth]{img/svm/rbf_kernel_21.pdf}& 6 | \includegraphics[width=0.4\textwidth]{img/svm/rbf_kernel_22.pdf} 7 | \end{tabular} 8 | \end{figure} 9 | \end{frame} -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/src/svm_example_linear_1.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{SVM example} 2 | Using a linear kernel: 3 | \begin{figure} 4 | \begin{tabular}{cc} 5 | \includegraphics[width=0.4\textwidth]{img/svm/linear_kernel_11.pdf}& 6 | \includegraphics[width=0.4\textwidth]{img/svm/linear_kernel_12.pdf} 7 | \end{tabular} 8 | \end{figure} 9 | \end{frame} -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/src/svm_example_linear_2.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{SVM example (2)} 2 | Using a linear kernel: 3 | \begin{figure} 4 | \begin{tabular}{cc} 5 | \includegraphics[width=0.4\textwidth]{img/svm/linear_kernel_21.pdf}& 6 | \includegraphics[width=0.4\textwidth]{img/svm/linear_kernel_22.pdf} 7 | \end{tabular} 8 | \end{figure} 9 | \end{frame} -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/src/svm_hard_margin.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{SVM: hard margin optimization} 2 | It can be shown that solving this constrained optimization problem maximizes the margin: 3 | \begin{equation*} 4 | \begin{aligned} 5 | & \min_{\vec{w},b} & & \frac{1}{2}||\vec{w}||^2\\ 6 | & \text{s.t.} & & y_i(\langle\vec{w},\vec{x_i}\rangle + b)>1 & \forall i=1,\ldots,N 7 | \end{aligned} 8 | \end{equation*} 9 | After estimating $\vec{w}$ and $b$, the decision rule for an unknown example $\vec{u}$ is simply: 10 | \begin{equation*} 11 | f(\vec{u}) = \text{sign}(\langle\vec{w},\vec{u}\rangle + b) 12 | \end{equation*} 13 | 14 | 15 | \end{frame} -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/src/svm_intro.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Support Vector Machines (SVM)} 2 | Very famous supervised learning algorithm: 3 | \begin{itemize} 4 | \item performs \textbf{binary} classification (can be adapted for multiclass problems) 5 | \item is \textbf{linear} (in its original formulation) 6 | \item main feature: chooses the decision boundary that \underline{maximizes the margin} between positive and negative examples 7 | \end{itemize} 8 | \begin{columns} 9 | \begin{column}{0.5\textwidth} 10 | \begin{center} 11 | \includegraphics[width=0.7\textwidth]{img/svm/max_margin.pdf} 12 | \end{center} 13 | \end{column} 14 | \begin{column}{0.5\textwidth} 15 | How do you find such line? 16 | \end{column} 17 | \end{columns} 18 | \end{frame} -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/src/svm_non_linear.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{SVM: non separable data} 2 | What if you cannot find a linear decision boundary? 3 | \begin{center} 4 | \includegraphics[width=0.35\textwidth]{img/svm/non_linear_data.pdf} 5 | \end{center} 6 | Two solutions: 7 | \begin{itemize} 8 | \item introduce slack variables $\xi_i$ (patch solution) 9 | \item kernel trick 10 | \end{itemize} 11 | \end{frame} -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/src/svm_setting.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{SVM: problem setting} 2 | \begin{itemize} 3 | \item we have a set of examples $\{\vec{x_i}\}_{i=1}^N$, with label $\{y_i\}_{i=1}^N$ 4 | \begin{itemize} 5 | \item $\vec{x_i}$ is a feature vector 6 | \item $y_i \in \{-1, 1\} \quad \forall i=1,\ldots,N$ 7 | \end{itemize} 8 | \item the line that maximizes the margin (hyperplane) is identified by 9 | \begin{itemize} 10 | \item the vector $\vec{w}$ orthogonal to it 11 | \item the intercept $b$ 12 | \end{itemize} 13 | \end{itemize} 14 | \end{frame} -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/src/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/support_vector_machines.dvi: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/support_vector_machines.dvi -------------------------------------------------------------------------------- /lab/support_vector_machines/tex/support_vector_machines.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/support_vector_machines/tex/support_vector_machines.pdf -------------------------------------------------------------------------------- /lab/tensorflow_convolutional_nets/README.md: -------------------------------------------------------------------------------- 1 | ### Lab lecture - Convolutional Neural Networks 2 | In this lecture we implement a CNN in Tensorflow, in the context 3 | of digits recognition (MNIST) and tiles segmentation (synthetic dataset). -------------------------------------------------------------------------------- /lab/tensorflow_convolutional_nets/tex/bibliography.bib: -------------------------------------------------------------------------------- 1 | %%%% MNIST 2 | @article{lecun1998mnist, 3 | title={The MNIST database of handwritten digits}, 4 | author={LeCun, Yann}, 5 | journal={http://yann. lecun. com/exdb/mnist/}, 6 | year={1998} 7 | } 8 | -------------------------------------------------------------------------------- /lab/tensorflow_convolutional_nets/tex/common/outline.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Agenda} 2 | \tableofcontents 3 | \end{frame} 4 | -------------------------------------------------------------------------------- /lab/tensorflow_convolutional_nets/tex/common/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /lab/tensorflow_convolutional_nets/tex/img/template/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_convolutional_nets/tex/img/template/logo_unimore_white.png -------------------------------------------------------------------------------- /lab/tensorflow_convolutional_nets/tex/img/tf/mnist.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_convolutional_nets/tex/img/tf/mnist.png -------------------------------------------------------------------------------- /lab/tensorflow_convolutional_nets/tex/img/tf/tensorboard.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_convolutional_nets/tex/img/tf/tensorboard.jpg -------------------------------------------------------------------------------- /lab/tensorflow_convolutional_nets/tex/img/tf/tensorboard_tiles.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_convolutional_nets/tex/img/tf/tensorboard_tiles.png -------------------------------------------------------------------------------- /lab/tensorflow_convolutional_nets/tex/img/tf/tiles_dataset.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_convolutional_nets/tex/img/tf/tiles_dataset.png -------------------------------------------------------------------------------- /lab/tensorflow_convolutional_nets/tex/tensorflow_convnets.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_convolutional_nets/tex/tensorflow_convnets.pdf -------------------------------------------------------------------------------- /lab/tensorflow_deep_autoencoders/README.md: -------------------------------------------------------------------------------- 1 | ### Lab lecture - Semi-supervised learning in Tensorflow 2 | This lecture is about implementing a deep denoising autoencoder (DAE) for learning a robust representation, the latter being suitable for a classifier when just few labeled data are available. 3 | ![Alt text](tex/img/autoencoder/semi.png?raw=true "Semi-supervised Learning with DAE") -------------------------------------------------------------------------------- /lab/tensorflow_deep_autoencoders/tex/Deep_Autoencoders.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_deep_autoencoders/tex/Deep_Autoencoders.pdf -------------------------------------------------------------------------------- /lab/tensorflow_deep_autoencoders/tex/common/outline.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Agenda} 2 | \tableofcontents 3 | \end{frame} 4 | -------------------------------------------------------------------------------- /lab/tensorflow_deep_autoencoders/tex/common/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /lab/tensorflow_deep_autoencoders/tex/img/autoencoder/accuracies.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_deep_autoencoders/tex/img/autoencoder/accuracies.png -------------------------------------------------------------------------------- /lab/tensorflow_deep_autoencoders/tex/img/autoencoder/denoising.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_deep_autoencoders/tex/img/autoencoder/denoising.pdf -------------------------------------------------------------------------------- /lab/tensorflow_deep_autoencoders/tex/img/autoencoder/mnist_noisy.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_deep_autoencoders/tex/img/autoencoder/mnist_noisy.png -------------------------------------------------------------------------------- /lab/tensorflow_deep_autoencoders/tex/img/autoencoder/naive.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_deep_autoencoders/tex/img/autoencoder/naive.pdf -------------------------------------------------------------------------------- /lab/tensorflow_deep_autoencoders/tex/img/autoencoder/reconstructions.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_deep_autoencoders/tex/img/autoencoder/reconstructions.png -------------------------------------------------------------------------------- /lab/tensorflow_deep_autoencoders/tex/img/autoencoder/semi.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_deep_autoencoders/tex/img/autoencoder/semi.pdf -------------------------------------------------------------------------------- /lab/tensorflow_deep_autoencoders/tex/img/autoencoder/semi.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_deep_autoencoders/tex/img/autoencoder/semi.png -------------------------------------------------------------------------------- /lab/tensorflow_deep_autoencoders/tex/img/template/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_deep_autoencoders/tex/img/template/logo_unimore_white.png -------------------------------------------------------------------------------- /lab/tensorflow_introduction/README.md: -------------------------------------------------------------------------------- 1 | ### Lab lecture - Tensorflow introduction 2 | This lecture is a brief introduction to Tensorflow programming, 3 | and does not involve implementation. -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/common/outline.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Agenda} 2 | \tableofcontents 3 | \end{frame} 4 | -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/common/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/img/template/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_introduction/tex/img/template/logo_unimore_white.png -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/img/tf/computational_graph.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_introduction/tex/img/tf/computational_graph.png -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/img/tf/graph_definition_0.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_introduction/tex/img/tf/graph_definition_0.jpg -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/img/tf/graph_definition_1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_introduction/tex/img/tf/graph_definition_1.jpg -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/img/tf/graph_evaluation.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_introduction/tex/img/tf/graph_evaluation.jpg -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/img/tf/placeholders.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_introduction/tex/img/tf/placeholders.png -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/img/tf/restore_variables.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_introduction/tex/img/tf/restore_variables.png -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/img/tf/save_variables.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_introduction/tex/img/tf/save_variables.png -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/img/tf/tensorboard.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_introduction/tex/img/tf/tensorboard.jpg -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/img/tf/tf_logo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_introduction/tex/img/tf/tf_logo.png -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/img/tf/variable_creation.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_introduction/tex/img/tf/variable_creation.jpg -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/img/tf/variable_initialization.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_introduction/tex/img/tf/variable_initialization.png -------------------------------------------------------------------------------- /lab/tensorflow_introduction/tex/tensorflow_intro.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_introduction/tex/tensorflow_intro.pdf -------------------------------------------------------------------------------- /lab/tensorflow_linear_regression/README.md: -------------------------------------------------------------------------------- 1 | ### Lab lecture - Linear regression in Tensorflow 2 | This lecture is about implementing a linear regression and train 3 | it by gradient descent, exploiting the automatic differentiation 4 | engine provided by Tensorflow. We'll see how to discover the relation 5 | between the weight of brain and body for a number of mammal species. -------------------------------------------------------------------------------- /lab/tensorflow_linear_regression/tex/common/outline.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Agenda} 2 | \tableofcontents 3 | \end{frame} 4 | -------------------------------------------------------------------------------- /lab/tensorflow_linear_regression/tex/common/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /lab/tensorflow_linear_regression/tex/img/template/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_linear_regression/tex/img/template/logo_unimore_white.png -------------------------------------------------------------------------------- /lab/tensorflow_linear_regression/tex/tensorflow_regression.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_linear_regression/tex/tensorflow_regression.pdf -------------------------------------------------------------------------------- /lab/tensorflow_neural_network/README.md: -------------------------------------------------------------------------------- 1 | ### Lab lecture - Neural Networks in Tensorflow 2 | In this lecture we employ Tensorflow to build a Multi-Layer Perceptron 3 | for the purpose of digits classification (MNIST). -------------------------------------------------------------------------------- /lab/tensorflow_neural_network/tex/bibliography.bib: -------------------------------------------------------------------------------- 1 | %%%% MNIST 2 | @article{lecun1998mnist, 3 | title={The MNIST database of handwritten digits}, 4 | author={LeCun, Yann}, 5 | journal={http://yann. lecun. com/exdb/mnist/}, 6 | year={1998} 7 | } 8 | -------------------------------------------------------------------------------- /lab/tensorflow_neural_network/tex/common/outline.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Agenda} 2 | \tableofcontents 3 | \end{frame} 4 | -------------------------------------------------------------------------------- /lab/tensorflow_neural_network/tex/common/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /lab/tensorflow_neural_network/tex/img/template/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_neural_network/tex/img/template/logo_unimore_white.png -------------------------------------------------------------------------------- /lab/tensorflow_neural_network/tex/img/tf/mnist.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_neural_network/tex/img/tf/mnist.png -------------------------------------------------------------------------------- /lab/tensorflow_neural_network/tex/tensorflow_neural_nets.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_neural_network/tex/tensorflow_neural_nets.pdf -------------------------------------------------------------------------------- /lab/tensorflow_recurrent_nets/README.md: -------------------------------------------------------------------------------- 1 | ### Lab lecture - Recurrent Networks in Tensorflow 2 | In this lecture we implement a Long-Short Term Memory (LSTM) 3 | in Tensorflow, and use it to learn to count the number of ones 4 | in a binary sequence. -------------------------------------------------------------------------------- /lab/tensorflow_recurrent_nets/tex/bibliography.bib: -------------------------------------------------------------------------------- 1 | %%%% LSTM 2 | @article{hochreiter1997long, 3 | title={Long short-term memory}, 4 | author={Hochreiter, Sepp and Schmidhuber, J{\"u}rgen}, 5 | journal={Neural computation}, 6 | volume={9}, 7 | number={8}, 8 | pages={1735--1780}, 9 | year={1997}, 10 | publisher={MIT Press} 11 | } 12 | -------------------------------------------------------------------------------- /lab/tensorflow_recurrent_nets/tex/common/outline.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Agenda} 2 | \tableofcontents 3 | \end{frame} 4 | -------------------------------------------------------------------------------- /lab/tensorflow_recurrent_nets/tex/common/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /lab/tensorflow_recurrent_nets/tex/img/template/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_recurrent_nets/tex/img/template/logo_unimore_white.png -------------------------------------------------------------------------------- /lab/tensorflow_recurrent_nets/tex/tensorflow_lstm.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/lab/tensorflow_recurrent_nets/tex/tensorflow_lstm.pdf -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/common/outline.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Agenda} 2 | \tableofcontents 3 | \end{frame} 4 | -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/common/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/convolutional_neural_networks.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/convolutional_neural_networks.pdf -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/act_relu.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/act_relu.jpg -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/act_sigmoid.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/act_sigmoid.jpg -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/act_tanh.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/act_tanh.jpg -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/activation_maps.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/activation_maps.jpg -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/black_box.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/black_box.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/busted_0.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/busted_0.pdf -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/busted_1.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/busted_1.pdf -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/cnn_architecture.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/cnn_architecture.jpg -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/cnn_vs_dnn.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/cnn_vs_dnn.jpg -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-0.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-0.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-1.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-10.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-10.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-11.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-11.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-12.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-12.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-13.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-13.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-14.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-14.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-15.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-15.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-16.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-16.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-17.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-17.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-18.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-18.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-19.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-19.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-2.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-20.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-20.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-21.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-21.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-22.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-22.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-23.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-23.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-24.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-24.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-3.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-4.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-5.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-6.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-7.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-8.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-8.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-9.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation/conv_animation-9.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation_stride/conv_animation_stride-0.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation_stride/conv_animation_stride-0.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation_stride/conv_animation_stride-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation_stride/conv_animation_stride-1.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation_stride/conv_animation_stride-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation_stride/conv_animation_stride-2.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation_stride/conv_animation_stride-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/conv_animation_stride/conv_animation_stride-3.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/convolution_gif.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/convolution_gif.gif -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/convolution_gif_stride: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/convolution_gif_stride -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/convolution_gif_stride_2.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/convolution_gif_stride_2.gif -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/finetuning_rules.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/finetuning_rules.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/learned_weights.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/learned_weights.jpg -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/local_connectivity.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/local_connectivity.jpg -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/maxpool.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/maxpool.jpg -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/occlusions.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/occlusions.jpg -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/pool.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/pool.jpg -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/semseg_in.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/semseg_in.jpg -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/semseg_out.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/semseg_out.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/training_possibilities.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/training_possibilities.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/tsne.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/tsne.jpg -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/vgg16_architecture.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/vgg16_architecture.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/visualize_activation.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/visualize_activation.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/cnn/weights_inspection.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/cnn/weights_inspection.png -------------------------------------------------------------------------------- /slides/deep_learning/convolutional_neural_networks/img/template/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/convolutional_neural_networks/img/template/logo_unimore_white.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/common/outline.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Agenda} 2 | \tableofcontents 3 | \end{frame} 4 | -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/common/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/deep_neural_networks.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/deep_neural_networks.pdf -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/2003_large_dataset.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/2003_large_dataset.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/2017_large_dataset.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/2017_large_dataset.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/act_relu.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/act_relu.jpg -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/act_sigmoid.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/act_sigmoid.jpg -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/act_tanh.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/act_tanh.jpg -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/augmentation.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/augmentation.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop.gif -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_0.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_0.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_1.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_2.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_3.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_4.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_5.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_6.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_7.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_8.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/backprop/backprop_8.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/biological_neuron.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/biological_neuron.jpg -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/cifar10.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/cifar10.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/deep_learning_media.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/deep_learning_media.jpg -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/deep_learning_scale.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/deep_learning_scale.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/dnn_overfit.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/dnn_overfit.jpg -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/dnn_overfit_reg.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/dnn_overfit_reg.jpg -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/dropout_no.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/dropout_no.jpg -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/dropout_yes.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/dropout_yes.jpg -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/forward_prop.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/forward_prop.jpg -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/linear_classifier.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/linear_classifier.jpg -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/log_plot.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/log_plot.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/neural_network.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/neural_network.jpg -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/overfitting.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/overfitting.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/single_neuron.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/single_neuron.jpg -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/single_neuron_backward.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/single_neuron_backward.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/dnn/single_neuron_forward.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/dnn/single_neuron_forward.png -------------------------------------------------------------------------------- /slides/deep_learning/deep_neural_networks/img/template/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/deep_neural_networks/img/template/logo_unimore_white.png -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/common/outline.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Agenda} 2 | \tableofcontents 3 | \end{frame} 4 | -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/common/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/gradient_descent.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/gradient_descent.pdf -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/convex_function.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/convex_function.png -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/descent/descent-0.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/descent/descent-0.png -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/descent/descent-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/descent/descent-1.png -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/descent/descent-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/descent/descent-2.png -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/descent/descent-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/descent/descent-3.png -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/descent/descent-4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/descent/descent-4.png -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/descent/descent-5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/descent/descent-5.png -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/descent/descent-6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/descent/descent-6.png -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/descent/descent-7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/descent/descent-7.png -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/level_sets.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/level_sets.png -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/lr_too_big.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/lr_too_big.jpg -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/lr_too_big.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/lr_too_big.png -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/lr_too_small.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/lr_too_small.jpg -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/lr_too_small.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/lr_too_small.png -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/sgd/non_convex_function.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/sgd/non_convex_function.jpg -------------------------------------------------------------------------------- /slides/deep_learning/gradient_descent/img/template/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/gradient_descent/img/template/logo_unimore_white.png -------------------------------------------------------------------------------- /slides/deep_learning/recurrent_neural_networks/common/outline.tex: -------------------------------------------------------------------------------- 1 | \begin{frame}{Agenda} 2 | \tableofcontents 3 | \end{frame} 4 | -------------------------------------------------------------------------------- /slides/deep_learning/recurrent_neural_networks/common/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /slides/deep_learning/recurrent_neural_networks/img/rnn/bptt.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/recurrent_neural_networks/img/rnn/bptt.png -------------------------------------------------------------------------------- /slides/deep_learning/recurrent_neural_networks/img/rnn/bptt_unroll.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/recurrent_neural_networks/img/rnn/bptt_unroll.png -------------------------------------------------------------------------------- /slides/deep_learning/recurrent_neural_networks/img/rnn/lstm_cell.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/recurrent_neural_networks/img/rnn/lstm_cell.png -------------------------------------------------------------------------------- /slides/deep_learning/recurrent_neural_networks/img/rnn/rnn_api.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/recurrent_neural_networks/img/rnn/rnn_api.png -------------------------------------------------------------------------------- /slides/deep_learning/recurrent_neural_networks/img/rnn/rnn_unfold.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/recurrent_neural_networks/img/rnn/rnn_unfold.png -------------------------------------------------------------------------------- /slides/deep_learning/recurrent_neural_networks/img/rnn/vanilla_rnn.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/recurrent_neural_networks/img/rnn/vanilla_rnn.png -------------------------------------------------------------------------------- /slides/deep_learning/recurrent_neural_networks/img/rnn/vanilla_rnn_no_output.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/recurrent_neural_networks/img/rnn/vanilla_rnn_no_output.png -------------------------------------------------------------------------------- /slides/deep_learning/recurrent_neural_networks/img/template/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/recurrent_neural_networks/img/template/logo_unimore_white.png -------------------------------------------------------------------------------- /slides/deep_learning/recurrent_neural_networks/recurrent_neural_networks.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/deep_learning/recurrent_neural_networks/recurrent_neural_networks.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/01_introduction_and_model_free_learning.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/01_introduction_and_model_free_learning.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/agent_and_environment.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/agent_and_environment.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/blackjack.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/blackjack.jpg -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/blackjack2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/blackjack2.jpg -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/blackjack_mc_control.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/blackjack_mc_control.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/blackjack_value.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/blackjack_value.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/doors.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/doors.jpg -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/go_home_example.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/go_home_example.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/maze_model.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/maze_model.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/maze_policy.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/maze_policy.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/maze_setup.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/maze_setup.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/maze_value.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/maze_value.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/mc_control.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/mc_control.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/mc_policy_iteration.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/mc_policy_iteration.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/ml_taxonomy.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/ml_taxonomy.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/rl_taxonomy.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/rl_taxonomy.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/sarsa.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/sarsa.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/sarsa_control.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/sarsa_control.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/sarsa_windy_gridworld.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/sarsa_windy_gridworld.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/img/windy_gridworld.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/img/windy_gridworld.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/01_introduction_and_model_free_learning/logo_unimore_white.png -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/acknowledge.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Disclaimer} 3 | All this material is a free re-arrangement of \href{http://www0.cs.ucl.ac.uk/staff/d.silver/web/Teaching.html}{David Silver's UCL Course on RL}. 4 | \\ 5 | You are also encouraged to take a look to his \href{https://www.youtube.com/watch?v=2pWv7GOvuf0}{Youtube lectures}. 6 | \end{frame} 7 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/agent_components.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Inside a reinforcement learning agent} 3 | \begin{itemize} 4 | \item An agent may include one or more of these components: 5 | \begin{itemize} 6 | \item Policy: agent's behaviour function 7 | \item Value function: how good is each state and/or action 8 | \item Model: representation of the environment 9 | \end{itemize} 10 | \end{itemize} 11 | \end{frame} 12 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/bellman_equation_2.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Bellman Expectation Equation} 3 | The state-value function can again be decomposed into immediate reward plus discounted value of successor state, 4 | \begin{equation*} 5 | v_{\pi}(s) = \mathbb{E}_{\pi}[R_{t+1} + \gamma v_{\pi}(S_{t+1}) | S_t = s] 6 | \end{equation*} 7 | The action-value function can similarly be decomposed, 8 | \begin{equation*} 9 | q_{\pi}(s, a) = \mathbb{E}_{\pi}[R_{t+1} + \gamma q_{\pi}(S_{t+1}, A_{t+1}) | S_t = s, A_t = a] 10 | \end{equation*} 11 | \end{frame} 12 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/blackjack_mc_control.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Monte-Carlo control in blackjack} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.7\textwidth]{img/blackjack_mc_control.pdf} 6 | \end{figure} 7 | \end{frame} 8 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/blackjack_reprise.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Back to the blackjack example} 3 | \centering 4 | \includegraphics[width=0.75\textwidth]{img/blackjack2.jpg} 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/characteristics.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{RL Characteristics} 3 | What makes reinforcement learning different from other machine learning paradigms? 4 | \begin{itemize} 5 | \item There is no supervisor, only a \emph{reward} signal. 6 | \item Feedback is delayed, not instantaneous 7 | \item Time really matters (sequential, non i.i.d. data) 8 | \item Agent is \emph{active}: its actions affect the environment he lives in. 9 | \end{itemize} 10 | \end{frame} 11 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/empty_slide.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{} 3 | \end{frame} 4 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/epsilon_greedy.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{$\epsilon$-greedy exploration} 3 | \begin{itemize} 4 | \item Simplest idea for ensuring continual exploration 5 | \item All $m$ actions are tried with non-zero probability 6 | \item With probability 1 − $\epsilon$ choose the greedy action 7 | \item With probability $\epsilon$ choose an action at random 8 | \end{itemize} 9 | \begin{equation*} 10 | \pi(a|s) = \begin{cases} 11 | \epsilon / m + 1 - \epsilon,& \text{if } a^{\ast} = \argmax_{a \in \mathcal{A}} Q(s,a)\\ 12 | \epsilon / m, & \text{otherwise} 13 | \end{cases} 14 | \end{equation*} 15 | \end{frame} 16 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/go_home_example.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Driving home example: MC vs. TD} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.9\textwidth]{img/go_home_example.pdf} 6 | \end{figure} 7 | \end{frame} 8 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/maze_environment.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Maze example} 3 | \begin{minipage}{0.45\textwidth} 4 | \begin{figure} 5 | \centering 6 | \includegraphics[width=\textwidth]{img/maze_setup.pdf} 7 | \end{figure} 8 | \end{minipage} 9 | \begin{minipage}{0.45\textwidth} 10 | \begin{itemize} 11 | \item Rewards: -1 per time-step 12 | \item Actions: N, S, W, E 13 | \item States: Agent's location 14 | \end{itemize} 15 | \end{minipage} 16 | \end{frame} 17 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/maze_model.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Maze example: model} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.5\textwidth]{img/maze_model.pdf} 6 | \end{figure} 7 | \begin{itemize} 8 | \item Grid layout represent transition model $\mathcal{P}_{ss'}^a$ 9 | \item Numbers represent immediate reward $R_s^a$ from each state $s$ (same for all a) 10 | \end{itemize} 11 | \end{frame} 12 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/maze_policy.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Maze example: policy} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.5\textwidth]{img/maze_policy.pdf} 6 | \end{figure} 7 | \begin{itemize} 8 | \item Arrows represent policy $\pi(s)$ for each state $s$ 9 | \end{itemize} 10 | \end{frame} 11 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/maze_value_function.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Maze example: value function} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.5\textwidth]{img/maze_value.pdf} 6 | \end{figure} 7 | \begin{itemize} 8 | \item Numbers represent policy $v_{\pi}(s)$ for each state $s$ 9 | \end{itemize} 10 | \end{frame} 11 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/mc_algo.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Every-Visit Monte-Carlo Policy Evaluation} 3 | \begin{itemize} 4 | \item To evaluate state $s$ 5 | \item Every time-step $t$ that state $s$ is visited in an episode, 6 | \item Increment counter $N(s) \leftarrow N(s) + 1$ 7 | \item Increment total return $S(s) \leftarrow S(s) + Gt$ 8 | \item Value is estimated by mean return $V(s) = S(s)/N(s)$ 9 | \item By law of large numbers, $V(s) \rightarrow v_{\pi}(s)$ as $N(s) \rightarrow \infty$ 10 | \end{itemize} 11 | \end{frame} 12 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/mc_algo_incremental.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Incremental Monte-Carlo Updates} 3 | \begin{itemize} 4 | \item Update $V(s)$ incrementally after episode $S_1, A_1, R_2, ..., S_T$ 5 | \item Compute return $G_t$ 6 | \item For each state $S_t$ with return $G_t$ 7 | \begin{align*} 8 | &N(S_t) \leftarrow N(S_t) + 1 \\ 9 | &V(S_t) \leftarrow V(S_t) + \frac{1}{N(S_t)}(G_t − V(S_t)) 10 | \end{align*} 11 | \item Usually a running mean is employed, i.e. forget old episodes 12 | \begin{equation*} 13 | V(S_t) \leftarrow V(S_t) + \alpha(G_t − V(S_t)) 14 | \end{equation*} 15 | \end{itemize} 16 | \end{frame} 17 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/mc_blackjack_value.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Blackjack Value Function after Monte-Carlo Learning} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.657\textwidth]{img/blackjack_value.pdf} 6 | \end{figure} 7 | Policy: \highlight{stick} if sum of cards $\leq$20, otherwise \highlight{twist} 8 | \end{frame} 9 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/mc_control.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Monte-Carlo control} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.5\textwidth]{img/mc_control.pdf} 6 | \end{figure} 7 | \highlight{Every episode:}\\ 8 | \highlight{Policy evaluation} Monte-Carlo policy evaluation, $Q=q_{\pi}$\\ 9 | \highlight{Policy improvement} $\epsilon$-greedy policy improvement 10 | \end{frame} 11 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/mc_intro.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Monte-Carlo reinforcement learning} 3 | \begin{itemize} 4 | \item MC methods learn directly from episodes of experience 5 | \item MC is \emph{model-free}: no explicit knowledge of environment mechanisms 6 | \item MC learns from complete episodes 7 | \begin{itemize} 8 | \item Caveat: can only apply to \emph{episodic} environments (all episodes must terminate). 9 | \end{itemize} 10 | \item MC uses the simpliest possible idea: value = mean return 11 | \end{itemize} 12 | \end{frame} 13 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/mc_policy_iteration.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Monte-Carlo policy iteration} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.5\textwidth]{img/mc_policy_iteration.pdf} 6 | \end{figure} 7 | \highlight{Policy evaluation} Monte-Carlo policy evaluation, $Q=q_{\pi}$\\ 8 | \highlight{Policy improvement} \highlight{$\epsilon$}-greedy policy improvement 9 | \end{frame} 10 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/mc_vs_td_control.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{MC vs TD control} 3 | \begin{itemize} 4 | \item Temporal-difference (TD) learning has several advantages over Monte-Carlo (MC) 5 | \begin{itemize} 6 | \item Lower variance 7 | \item Online 8 | \item Incomplete sequences 9 | \end{itemize} 10 | \item Natural idea: use TD instead of MC in our control loop 11 | \begin{itemize} 12 | \item Apply TD to $Q(S, A)$ 13 | \item Use $\epsilon$-greedy policy improvement 14 | \item Update every time-step 15 | \end{itemize} 16 | \end{itemize} 17 | \end{frame} 18 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/ml_taxonomy.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Branches of Machine Learning} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.6\textwidth]{img/ml_taxonomy.pdf} 6 | \end{figure} 7 | \end{frame} 8 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/model.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Model} 3 | \begin{itemize} 4 | \item A \highlight{model} predicts what the environment will do next 5 | \item $\mathcal{P}$ predicts the next state 6 | \item $\mathcal{R}$ predicts the next (immediate) reward 7 | \end{itemize} 8 | \begin{equation*} 9 | \mathcal{P}^a{ss'}=\mathbb{P}[S_{t+1}=s'| S_t=s, A_t=a] 10 | \end{equation*} 11 | \begin{equation*} 12 | \mathcal{R}^a_s = \mathbb{E}[R_{t+1}|S_t=s, A_t=a] 13 | \end{equation*} 14 | \end{frame} 15 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/model_free_prediction.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Model-Free prediction} 3 | \begin{itemize} 4 | \item \highlight{Model-free prediction} 5 | \begin{itemize} 6 | \item estimate the value function given a policy in a non-observable environment 7 | \begin{itemize} 8 | \item Monte-Carlo Learning 9 | \item Temporal-Difference Learning 10 | \end{itemize} 11 | \end{itemize} 12 | \end{itemize} 13 | \end{frame} 14 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/on_off_policy.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{On and off policy learning} 3 | \begin{itemize} 4 | \item \highlight{On-policy learning} 5 | \begin{itemize} 6 | \item ``Learn on the job'' 7 | \item Learn about policy $\pi$ from experience sampled from $\pi$ 8 | \end{itemize} 9 | \item \highlight{Off-policy learning} 10 | \begin{itemize} 11 | \item ``Look over someone’s shoulder'' 12 | \item Learn about policy $\pi$ from experience sampled from $\mu$ 13 | \end{itemize} 14 | \end{itemize} 15 | \end{frame} 16 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/policy.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Policy} 3 | \begin{itemize} 4 | \item A \highlight{policy} is the agent's behaviour 5 | \item It is a map from state to action 6 | \item Deterministic policy: $a = \pi(s)$ 7 | \item Stochastic policy: $\pi(a|s) = \mathbb{P}[A_t=s | S_t=s]$ 8 | \end{itemize} 9 | \end{frame} 10 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/q_learning.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Q-learning} 3 | \begin{itemize} 4 | \item We now consider off-policy learning of action-values $Q(s, a)$ 5 | \item Next action is chosen using behaviour policy $A_{t+1} \sim \mu(\cdot | S_t)$ 6 | \item But we consider alternative successor action $A^{\prime} \sim \pi(\cdot|S_t)$ 7 | \item And update $Q(S_t, A_t)$ towards value of alternative action 8 | \end{itemize} 9 | \begin{equation*} 10 | Q(S_t, A_t) \leftarrow Q(S_t, A_t) + \alpha(\highlight{R_{t+1} + 11 | Q(S_{t+1}, A^{\prime})} - Q(S_t, A_t)) 12 | \end{equation*} 13 | \end{frame} 14 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/q_learning_algorithm.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Q-learning algorithm for off-policy control} 3 | \begin{algorithmic} 4 | \STATE Initialize $Q(s,a), \forall s \in S, a \in \mathcal{A}(s)$, arbitrarily, and $Q(terminal-state, \dot)=0$ 5 | \FOR{each episode} 6 | \STATE Intialise $S$ 7 | \FOR {each step of episode} 8 | \STATE Choose $A$ from $S$ using policy derived from $Q$ (e.g., $\epsilon$-greedy) 9 | \STATE Take action $A$, observe $R$, $S^{\prime}$ 10 | \STATE $Q(S,A) \leftarrow Q(S,A) + \alpha (R + \gamma \max_{a^{\prime}}Q(S^{\prime}, a^{\prime}) - Q(S,A))$ 11 | \STATE $S \leftarrow S^{\prime}$ 12 | \ENDFOR 13 | \ENDFOR 14 | \end{algorithmic} 15 | \end{frame} 16 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/rewards.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Rewards} 3 | \begin{itemize} 4 | \item A \highlight{reward} $R_t$ is a scalar feedback signal 5 | \item Indicates how well agent is doing at step $t$ 6 | \item The agent's job is to maximise cumulative reward over an episode 7 | \end{itemize} 8 | \end{frame} 9 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/rl_taxonomy.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{RL taxonomy} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.65\textwidth]{img/rl_taxonomy.pdf} 6 | \end{figure} 7 | \end{frame} 8 | \end{frame} 9 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/sarsa.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Updating action-value functions with SARSA} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.2\textwidth]{img/sarsa.pdf} 6 | \end{figure} 7 | \begin{equation*} 8 | Q(S,A) \leftarrow Q(S,A) + \alpha (R + \gamma Q(S^{\prime}, A^{\prime}) - Q(S,A)) 9 | \end{equation*} 10 | \end{frame} 11 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/sarsa_control.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{On-policy control with SARSA} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.5\textwidth]{img/sarsa_control.pdf} 6 | \end{figure} 7 | Every \highlight{time-step}:\\ 8 | \highlight{Policy evaluation} with \highlight{Sarsa}, $Q\approx q_{\pi}$\\ 9 | \highlight{Policy improvement} with $\epsilon$-greedy policy improvement. 10 | \end{frame} 11 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/sarsa_windy_gridworld.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{SARSA on the Windy Gridworld} 3 | \begin{figure} 4 | \centering 5 | \hspace{-1cm}\includegraphics[width=0.8\textwidth]{img/sarsa_windy_gridworld.pdf} 6 | \end{figure} 7 | \end{frame} 8 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/sequential_decision_making.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Sequential Decision Making} 3 | \begin{itemize} 4 | \item Goal: \emph{select actions to maximise total future reward} 5 | \item Actions may have long term consequences 6 | \item Reward may be delayed 7 | \item It may be better to sacrifice immediate reward to gain more long-term reward. 8 | \begin{itemize} 9 | \item A financial investment may take months to mature 10 | \item Refuelling a helicopter now might prevent a crash in several hours 11 | \item Blocking opponent moves might help winning chances many moves from now 12 | \end{itemize} 13 | \end{itemize} 14 | \end{frame} 15 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/state.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{State} 3 | \begin{itemize} 4 | \item The \highlight{history} is the sequence of observations, actions, rewards 5 | \begin{equation*} 6 | H_t = O_1, R_1, A_1, \ldots, A_{t-1}, O_t, R_t 7 | \end{equation*} 8 | \item The \highlight{state} is the information used to determine what happens next. 9 | \begin{itemize} 10 | \item It is a function of the history: 11 | \begin{equation*} 12 | S_t = f(H_t) 13 | \end{equation*} 14 | \end{itemize} 15 | \end{itemize} 16 | \end{frame} 17 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/td_learning_intro.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Temporal-Difference learning} 3 | \begin{itemize} 4 | \item TD methods learn directly from episodes of experience 5 | \item TD is \emph{model-free}: no knowledge of MDP transitions / rewards 6 | \item TD learns from \emph{incomplete} episodes, by \emph{bootstrapping} 7 | \item TD updates a guess towards a guess 8 | \end{itemize} 9 | \end{frame} 10 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/td_vs_mc_2.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Advantages and disadvantages of MC vs. TD} 3 | \begin{itemize} 4 | \item TD can learn \emph{before} knowing the final outcome 5 | \begin{itemize} 6 | \item TD can learn online after every step 7 | \item MC must wait until end of episode before return is known 8 | \end{itemize} 9 | \item TD can learn without the final outcome 10 | \begin{itemize} 11 | \item TD can learn from incomplete sequences 12 | \item MC can only learn from complete sequences 13 | \item TD works in continuing (non-terminating) environments 14 | \item MC only works for episodic (terminating) environments 15 | \end{itemize} 16 | \end{itemize} 17 | \end{frame} 18 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/td_vs_mc_3.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Bias/variance trade-off} 3 | \begin{itemize} 4 | \item Return $G_t=R_{t+1}+R_{t+2}+\ldots+\gamma^{T-1}R_T$ is \emph{unbiased} estimate of $v_{\pi}(S_t)$ 5 | \item True TD target $R_{t+1}+v_{\pi}(S_{t+1})$ is unbiased estimate of $v_{\pi}(S_t)$ 6 | \item TD target $R_{t+1}+v(S_{t+1})$ is biased estimate of $v_{\pi}(S_t)$ 7 | \item TD target is much lower variance than the return: 8 | \begin{itemize} 9 | \item Return depends on many random actions, transitions, rewards 10 | \item TD target depends on one random action, transition, reward 11 | \end{itemize} 12 | \end{itemize} 13 | \end{frame} 14 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/01_introduction_and_model_free_learning/src/windy_gridworld.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Windy gridworld example} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.7\textwidth]{img/windy_gridworld.pdf} 6 | \end{figure} 7 | \begin{itemize} 8 | \item Reward = -1 per time step until reaching goal 9 | \item Undiscounted 10 | \end{itemize} 11 | \end{frame} 12 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/02_function_approximation.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/02_function_approximation/02_function_approximation.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/img/dqn.JPG: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/02_function_approximation/img/dqn.JPG -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/img/dqn.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/02_function_approximation/img/dqn.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/img/dqn_results.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/02_function_approximation/img/dqn_results.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/img/gd.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/02_function_approximation/img/gd.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/img/mountain_car.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/02_function_approximation/img/mountain_car.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/img/vfa_control.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/02_function_approximation/img/vfa_control.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/img/vfa_types.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/02_function_approximation/img/vfa_types.pdf -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/logo_unimore_white.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/slides/reinforcement_learning/02_function_approximation/logo_unimore_white.png -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/acknowledge.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Disclaimer} 3 | All this material is a free re-arrangement of \href{http://www0.cs.ucl.ac.uk/staff/d.silver/web/Teaching.html}{David Silver's UCL Course on RL}. 4 | \\ 5 | You are also encouraged to take a look to his \href{https://www.youtube.com/watch?v=2pWv7GOvuf0}{Youtube lectures}. 6 | \end{frame} 7 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/batch_rl.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Batch reinforcement learning} 3 | \begin{itemize} 4 | \item Gradient descent is simple and appealing 5 | \item But it is not sample efficient 6 | \item Batch methods seek to find the best fitting value function 7 | \item Given the agent's experience (``training data'') 8 | \end{itemize} 9 | \end{frame} 10 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/convergence_control.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Convergence of prediction algorithms} 3 | \begin{table} 4 | \begin{tabular}{cccc} 5 | \hline 6 | Algorithm & Table Lookup & Linear & Non-Linear\\ \hline 7 | Monte-Carlo Control & \cmark & (\cmark) & \xmark \\ 8 | Sarsa & \cmark & (\cmark) & \xmark \\ 9 | Q-learning & \cmark & \xmark & \xmark \\\hline 10 | \end{tabular} 11 | \end{table} 12 | (\cmark) = chatters around near-optimal value function 13 | \end{frame} 14 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/convergence_prediction.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Convergence of prediction algorithms} 3 | \begin{table} 4 | \begin{tabular}{ccccc} 5 | \hline 6 | On/Off Policy & Algorithm & Table Lookup & Linear & Non-Linear\\ \hline 7 | \multirow{2}{*}{On-policy} & MC & \cmark & \cmark & \cmark \\ 8 | & TD(0) & \cmark & \cmark & \xmark\\ \hline 9 | \multirow{2}{*}{Off-policy} & MC & \cmark & \cmark & \cmark \\ 10 | & TD(0) & \cmark & \xmark & \xmark\\ \hline 11 | 12 | \end{tabular} 13 | \end{table} 14 | \end{frame} 15 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/dqn_atari.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{DQN in Atari} 3 | \begin{itemize} 4 | \item End-to-end learning of values $Q(s, a)$ from pixels $s$ 5 | \item Input state $s$ is stack of raw pixels from last 4 frames 6 | \item Output is $Q(s, a)$ for 18 joystick/button positions 7 | \item Reward is change in score for that step 8 | \end{itemize} 9 | \begin{figure} 10 | \centering 11 | \includegraphics[width=0.8\textwidth]{img/dqn.JPG} 12 | \end{figure} 13 | \end{frame} 14 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/dqn_results.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{DQN results in Atari} 3 | \centering 4 | \includegraphics[width=0.85\textwidth]{img/dqn_results.pdf} 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/empty_slide.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{} 3 | \end{frame} 4 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/feature_vectors.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Feature vectors} 3 | \begin{itemize} 4 | \item Represent state by a feature vector 5 | \begin{equation*} 6 | \textbf{x}(S) = \left(\begin{array}{c} 7 | \textbf{x}_1(S)\\ 8 | \vdots\\ 9 | \textbf{x}_n(S) 10 | \end{array} 11 | \right) 12 | \end{equation*} 13 | \item For example: 14 | \begin{itemize} 15 | \item Distance of robot from landmarks 16 | \item Trends in the stock market 17 | \item Piece and pawn configurations in chess 18 | \end{itemize} 19 | \end{itemize} 20 | \end{frame} 21 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/large_scale_RL.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Large scale reinforcement learning} 3 | Reinforcement learning can be used to solve large problems, e.g. 4 | \begin{itemize} 5 | \item Backgammon: $10^{20}$ states 6 | \item Computer Go: $10^{170}$ states 7 | \item Helicopter: continuous state space 8 | \end{itemize} 9 | \onslide<2->{ 10 | How can we scale up the model-free methods for prediction and control from the last lecture?} 11 | \end{frame} 12 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/mountain_car.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Linear SARSA in mountain car} 3 | \centering 4 | \includegraphics[width=0.8\textwidth]{img/mountain_car} 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/titlepage.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \renewcommand{\insertframenumber}{} 3 | \begin{frame}[noframenumbering] 4 | \titlepage 5 | \end{frame} 6 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/vfa_control.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Control with value function approximation} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.6\textwidth]{img/vfa_control} 6 | \end{figure} 7 | \highlight{Policy evaluation Approximate} policy evaluation, $\hat{q}(\cdot, \cdot, \textbf{w}) \approx q_{\pi}$\\ 8 | \highlight{Policy improvement} $\epsilon$-greedy policy improvement 9 | \end{frame} 10 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/vfa_types.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Types of value function approximation} 3 | \begin{figure} 4 | \centering 5 | \includegraphics[width=0.8\textwidth]{img/vfa_types.pdf} 6 | \end{figure} 7 | \end{frame} 8 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/which_function.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Which function approximator?} 3 | There are many function approximators, e.g. 4 | \begin{itemize} 5 | \item Linear combinations of features 6 | \item Neural network 7 | \item Decision tree 8 | \item Nearest neighbour 9 | \item Fourier / wavelet bases 10 | \item \ldots 11 | \end{itemize} 12 | \end{frame} 13 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/02_function_approximation/src/which_function_2.tex: -------------------------------------------------------------------------------- 1 | \bgroup 2 | \begin{frame}{Which function approximator?} 3 | We consider \highlight{differentiable} function approximators, e.g.\begin{itemize} 4 | \item \highlight{Linear combinations of features} 5 | \item \highlight{Neural network} 6 | \item Decision tree 7 | \item Nearest neighbour 8 | \item Fourier / wavelet bases 9 | \item \ldots 10 | \end{itemize} 11 | Furthermore, we require a training method that is suitable for \highlight{non-stationary, non-iid} data 12 | \end{frame} 13 | \egroup -------------------------------------------------------------------------------- /slides/reinforcement_learning/README.md: -------------------------------------------------------------------------------- 1 | # reinforcement-learning-lectures -------------------------------------------------------------------------------- /thumbs/01_introduction_and_model_free_learning.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/01_introduction_and_model_free_learning.gif -------------------------------------------------------------------------------- /thumbs/02_function_approximation.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/02_function_approximation.gif -------------------------------------------------------------------------------- /thumbs/boosting.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/boosting.gif -------------------------------------------------------------------------------- /thumbs/clustering.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/clustering.gif -------------------------------------------------------------------------------- /thumbs/convolutional_neural_networks.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/convolutional_neural_networks.gif -------------------------------------------------------------------------------- /thumbs/deep_autoencoders.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/deep_autoencoders.gif -------------------------------------------------------------------------------- /thumbs/deep_neural_networks.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/deep_neural_networks.gif -------------------------------------------------------------------------------- /thumbs/dimensionality_reduction.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/dimensionality_reduction.gif -------------------------------------------------------------------------------- /thumbs/gradient_descent.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/gradient_descent.gif -------------------------------------------------------------------------------- /thumbs/logistic_regression.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/logistic_regression.gif -------------------------------------------------------------------------------- /thumbs/naive_bayes.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/naive_bayes.gif -------------------------------------------------------------------------------- /thumbs/q_learning.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/q_learning.gif -------------------------------------------------------------------------------- /thumbs/recurrent_neural_networks.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/recurrent_neural_networks.gif -------------------------------------------------------------------------------- /thumbs/support_vector_machines.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/support_vector_machines.gif -------------------------------------------------------------------------------- /thumbs/tensorflow_convnets.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/tensorflow_convnets.gif -------------------------------------------------------------------------------- /thumbs/tensorflow_intro.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/tensorflow_intro.gif -------------------------------------------------------------------------------- /thumbs/tensorflow_lstm.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/tensorflow_lstm.gif -------------------------------------------------------------------------------- /thumbs/tensorflow_neural_nets.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/tensorflow_neural_nets.gif -------------------------------------------------------------------------------- /thumbs/tensorflow_regression.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ndrplz/machine_learning_lectures/d332476e354a8560cfa8e8181bca5dd53b95342d/thumbs/tensorflow_regression.gif --------------------------------------------------------------------------------