├── 2017年10月 ├── 20170925 第1期 │ ├── README.md │ ├── 人工智能先驱说我们需要重新开始.md │ ├── 寻找Keras支持的最快的深度学习框架.md │ └── 深度学习目标检测：一个全面的回顾.md ├── 20171004 第2期 │ ├── README.md │ ├── 卷积神经网络的直观解释.md │ ├── 我在两个月前是如何开始学AI的.md │ └── 深度学习是否应使用复数？.md ├── 20171005 第3期 │ ├── README.md │ ├── git.md │ └── 神经网络入门介绍.md ├── 20171006 第4期 │ ├── Go语言十年回顾.md │ ├── README.md │ ├── 为什么我们开发TiKV选择Rust而不用Golang或者CC++.md │ └── 设计深度学习项目.md ├── 20171007 第5期 │ ├── README.md │ ├── 大脑vs深度学习的第一部分：计算复杂度（为什么奇点离我们还很远） .md │ └── 愚蠢的专利正在拖垮人工智能和机器学习.md ├── 20171008 第6期 │ ├── AI是否一招鲜吃遍天？.md │ ├── IOS中的计算机视觉——目标检测.md │ ├── README.md │ └── 神经网络生成虚假的创业公司，随之而来的欢乐.md ├── 20171009 第7期 │ ├── 5 Ways the Internet of Things Is Revolutionizing Healthcare.md │ ├── README.md │ └── Recurrent Neural Networks.md ├── 20171010 第8期 │ ├── ML算法附录：被动攻击算法.md │ ├── README.md │ └── 在深度学习中关于不同卷积的介绍.md ├── 20171011 第9期 │ ├── README.md │ ├── tensorflow.md │ └── 深度学习的局限性.md ├── 20171012 第10期 │ ├── README.md │ ├── 信息理论概论和你应该关心的原因.md │ ├── 工程工作流程研究.md │ └── 浅谈词袋模型.md ├── 20171013 第11期 │ ├── PyTorch与TensorFlow 一个月总结.md │ ├── README.md │ ├── 将卷积神经网络集成到企业级应用中.md │ └── 浅谈10个通用的软件架构模式.md ├── 20171014 第12期 │ ├── Go中的管道模式.md │ ├── README.md │ ├── 可视化信息论.md │ └── 管理人脸识别AI的时间还很长.md ├── 20171015 第13期 │ ├── Intro To Data Analysis For Everyone! Part 1.md │ ├── PyTorch tutorial distilled.md │ ├── README.md │ └── The Search for Better Search at Reddit.md ├── 20171016 第14期 │ ├── A Quick Introduction to Neural Networks.md │ ├── BUILDING A NEURAL NET FROM SCRATCH IN GO.md │ ├── Practical Data Science in Python.md │ ├── README.md │ └── 神经网络的简单介绍.md ├── 20171017 第15期 │ ├── MapReduce的模式、算法和用例.md │ ├── NoSQL数据库中的分布式算法.md │ ├── README.md │ └── 软件工程师必备的机器学习技能.md ├── 20171018 第16期 │ ├── Benefits of Intel® Optimized Caffe in comparison with BVLC Caffe.md │ ├── NEW OPTIMIZATIONS IMPROVE DEEP LEARNING FRAMEWORKS FOR CPUS.md │ ├── README.md │ └── Why SQL is beating NoSQL, and what this means for the future of data.md ├── 20171019 第17期 │ ├── README.md │ ├── TensorFlow在现代英特尔体系结构下的优化.md │ ├── 一个用Kears和 OpenAi Gym实现深度Q网络的Gotchas指南.md │ └── 科学家可以读取一只鸟的大脑并且预测它的下一声啼叫.md ├── 20171020 第18期 │ ├── AlphaGoZero从零开始学习.md │ ├── Perceptually Uniform Color Spaces.md │ ├── README.md │ └── Stop Using Word2vec.md ├── 20171021 第19期 │ ├── 8 Essential Tips for People starting a Career in Data Science.md │ ├── Hey Siri An On-device DNN-powered Voice Trigger for Apple’s Personal Assistant.md │ ├── How to Prepare Movie Review Data for Sentiment Analysis.md │ └── README.md ├── 20171022 第20期 │ ├── 11行python实现神经网络.md │ ├── AlphaGo与能力增强.md │ ├── README.md │ └── 使用YOLO提高实时目标检测能力.md ├── 20171023 第21期 │ ├── Bayesian Learning for Statistical Classification.md │ ├── Instacart Market Basket Analysis, Winner's Interview 2nd place, Kazuki Onodera.md │ ├── Intel® Nervana™ Neural Network Processors (NNP) Redefine AI Silicon.md │ └── README.md ├── 20171024 第22期 │ ├── Advice For New and Junior Data Scientists.md │ ├── README.md │ ├── Subword-level embeddings.md │ └── Word embeddings in 2017 Trends and future directions.md ├── 20171025 第23期 │ ├── README.md │ └── The difference between Statistical Modeling and Machine Learning, as I see it.md ├── 20171026 第24期 │ ├── Batch normalization in Neural Networks.md │ ├── Natural Stupidity is more Dangerous than Artificial Intelligence.md │ ├── README.md │ └── Why AlphaGo Zero is a Quantum Leap Forward in Deep Learning.md ├── 20171027 第25期 │ ├── Approaching (Almost) Any Machine Learning Problem Abhishek Thakur.md │ ├── Google Says Machine Learning Chips Make AI Faster and More Efficient.md │ ├── README.md │ └── Word Tensors.md ├── 20171028 第26期 │ ├── 10 tips on using Jupyter Notebook.md │ ├── Best Practices for Document Classification with Deep Learning.md │ ├── README.md │ └── 如何使用C++写出一个TensorFlow.md ├── 20171029 第27期 │ ├── Fundamentals of Deep Learning – Activation Functions and When to Use Them.md │ ├── Planet Understanding the Amazon from Space, 1st Place Winner's Interview.md │ ├── README.md │ └── Speech Recognition Is Not Solved.md ├── 20171030 第28期 │ └── README.md └── 20171031 第29期 │ └── README.md ├── 2017年11月 ├── 20171101 第30期 │ └── README.md ├── 20171102 第31期 │ └── README.md ├── 20171103 第32期 │ └── README.md ├── 20171104 第33期 │ └── README.md ├── 20171105 第34期 │ └── README.md ├── 20171106 第35期 │ └── README.md ├── 20171107 第36期 │ └── README.md ├── 20171108 第37期 │ └── README.md ├── 20171109 第38期 │ └── README.md ├── 20171110 第39期 │ └── README.md ├── 20171111 第40期 │ └── README.md ├── 20171112 第41期 │ └── README.md ├── 20171113 第42期 │ └── README.md ├── 20171114 第43期 │ └── README.md ├── 20171115 第44期 │ └── README.md ├── 20171116 第45期 │ └── README.md ├── 20171117 第46期 │ └── README.md ├── 20171118 第47期 │ └── README.md ├── 20171119 第48期 │ └── README.md ├── 20171120 第49期 │ └── README.md ├── 20171121 第50期 │ └── README.md ├── 20171122 第51期 │ └── README.md ├── 20171123 第52期 │ └── README.md ├── 20171124 第53期 │ └── README.md ├── 20171125 第54期 │ └── README.md ├── 20171126 第55期 │ └── README.md ├── 20171127 第56期 │ └── README.md ├── 20171128 第57期 │ └── README.md ├── 20171129 第58期 │ └── README.md └── 20171130 第59期 │ └── README.md ├── 2017年12月 ├── 20171201 第60期 │ └── README.md ├── 20171202 第61期 │ └── README.md ├── 20171203 第62期 │ └── README.md ├── 20171204 第63期 │ └── README.md ├── 20171205 第64期 │ └── README.md ├── 20171206 第65期 │ └── README.md ├── 20171207 第66期 │ └── README.md ├── 20171208 第67期 │ └── README.md ├── 20171209 第68期 │ └── README.md ├── 20171210 第69期 │ └── README.md ├── 20171211 第70期 │ └── README.md ├── 20171212 第71期 │ └── README.md ├── 20171213 第72期 │ └── README.md ├── 20171214 第73期 │ └── README.md ├── 20171215 第74期 │ └── README.md ├── 20171216 第75期 │ └── README.md ├── 20171217 第76期 │ └── README.md ├── 20171218 第77期 │ └── README.md ├── 20171219 第78期 │ └── README.md ├── 20171220 第79期 │ └── README.md ├── 20171221 第80期 │ └── README.md ├── 20171222 第81期 │ └── README.md ├── 20171223 第82期 │ └── README.md ├── 20171224 第83期 │ └── README.md ├── 20171225 第84期 │ └── README.md ├── 20171226 第85期 │ └── README.md ├── 20171227 第86期 │ └── README.md ├── 20171228 第87期 │ └── README.md ├── 20171229 第88期 │ └── README.md ├── 20171230 第89期 │ └── README.md └── 20171231 第90期 │ └── README.md ├── 2018年1月 ├── 20180101 第91期 │ └── README.md ├── 20180102 第92期 │ └── README.md ├── 20180103 第93期 │ ├── .md │ ├── 10-fearless-predictions-for-deep-learning-in-2018.md │ ├── README.md │ └── logistic-regression-vs-deep-neural-networks-david.md ├── 20180104 第94期 │ └── README.md ├── 20180105 第95期 │ ├── README.md │ └── nips-2017-policy-field-notes.md ├── 20180106 第96期 │ └── README.md ├── 20180107 第97期 │ └── README.md ├── 20180108 第98期 │ └── README.md ├── 20180109 第99期 │ ├── README.md │ └── meltdown.md ├── 20180110 第100期 │ └── README.md ├── 20180111 第101期 │ └── README.md ├── 20180112 第102期 │ ├── README.md │ └── the-8-neural-network-architectures-machine-learning-researchers-need-to-learn.md ├── 20180113 第103期 │ └── README.md ├── 20180114 第104期 │ └── README.md ├── 20180115 第105期 │ └── README.md ├── 20180116 第106期 │ ├── README.md │ └── first-time-with-kaggle-a-convnet-to-classify-toxic-comments-with-keras.md ├── 20180117 第107期 │ ├── README.md │ └── the-3-tricks-that-made-alphago-zero-work.md ├── 20180118 第108期 │ └── README.md ├── 20180119 第109期 │ └── README.md ├── 20180120 第110期 │ └── README.md ├── 20180121 第111期 │ └── README.md ├── 20180122 第112期 │ └── README.md ├── 20180123 第113期 │ ├── 8-deep-learning-best-practices-i-learned-about-in-2017.md │ └── README.md ├── 20180124 第114期 │ ├── 12-useful-things-to-know-about-machine-learning.md │ └── README.md ├── 20180125 第115期 │ ├── README.md │ └── artwork-personalization.md ├── 20180126 第116期 │ └── README.md ├── 20180127 第117期 │ ├── README.md │ └── benchmarking-tensorflow-performance-on-next-generation-gpus.md ├── 20180128 第118期 │ ├── README.md │ └── how-to-build-your-own-alphazero-ai-using-python-and-keras.md ├── 20180129 第119期 │ ├── README.md │ ├── ethics-in-machine-learning.md │ └── what-was-ai-in-2017.md ├── 20180130 第120期 │ └── README.md └── 20180131 第121期 │ └── README.md ├── 2018年2月 ├── 20180201 第122期 │ ├── README.md │ └── getting-the-most-of-xgboost-and-lightgbm-speed-compiler-cpu-pinning.md ├── 20180202 第123期 │ └── README.md ├── 20180203 第124期 │ └── README.md ├── 20180204 第125期 │ ├── README.md │ └── a-genetic-algorithm-scheduled-our-developer-conference-heres-what-i-learnt.md ├── 20180205 第126期 │ ├── README.md │ └── asking-the-right-questions-about-ai.md ├── 20180206 第127期 │ ├── README.md │ └── super-simple-end-to-end-test-of-keras-tensorflow-and-coreml.md ├── 20180207 第128期 │ └── README.md ├── 20180208 第129期 │ └── README.md ├── 20180209 第130期 │ ├── README.md │ └── how-we-grew-from-0-to-4-million-women-on-our-fashion-app-with-a-vertical-machine-learning-approach.md ├── 20180210 第131期 │ ├── README.md │ └── snapchats-filters-how-computer-vision-recognizes-your-face.md ├── 20180211 第132期 │ ├── README.md │ ├── ava-the-art-and-science-of-image-discovery-at-netflix.md │ └── from-research-to-practice.md ├── 20180212 第133期 │ └── README.md ├── 20180213 第134期 │ └── README.md ├── 20180214 第135期 │ ├── README.md │ └── otodscinos-the-root-cause-of-slow-neural-net-training.md ├── 20180215 第136期 │ └── README.md ├── 20180216 第137期 │ └── README.md ├── 20180217 第138期 │ ├── README.md │ └── a-deeper-understanding-of-nnets-part-3-lstm-and-gru.md ├── 20180218 第139期 │ └── README.md ├── 20180219 第140期 │ ├── README.md │ └── mltest-automatically-test-neural-network-models-in-one-function-call.md ├── 20180220 第141期 │ └── README.md ├── 20180221 第142期 │ └── README.md ├── 20180222 第143期 │ └── README.md ├── 20180223 第144期 │ ├── README.md │ └── open-machine-learning-course-topic-3-classification-decision-trees-and-k-nearest-neighbors.md ├── 20180224 第145期 │ └── README.md ├── 20180225 第146期 │ └── README.md ├── 20180226 第147期 │ ├── README.md │ └── how-to-create-a-machine-learning-framework-from-scratch-in-491-steps.md ├── 20180227 第148期 │ └── README.md └── 20180228 第149期 │ ├── 13-deep-learning-frameworks-for-natural-language-processing-in-python.md │ └── README.md ├── 2018年3月 ├── 20180301 第150期 │ ├── Numpy的70个操作.md │ └── README.md ├── 20180302 第151期 │ └── README.md ├── 20180303 第152期 │ └── README.md ├── 20180304 第153期 │ └── README.md ├── 20180305 第154期 │ └── README.md ├── 20180306 第155期 │ └── README.md ├── 20180307 第156期 │ └── README.md ├── 20180308 第157期 │ ├── README.md │ └── only-numpy-why-i-do-manual-back-propagation-implementing-multi-channel-layer-convolution-neural.md ├── 20180309 第158期 │ └── README.md ├── 20180310 第159期 │ ├── README.md │ └── its-a-no-brainer-deep-learning-for-brain-mr-images.md ├── 20180311 第160期 │ └── README.md ├── 20180312 第161期 │ └── README.md ├── 20180313 第162期 │ └── README.md ├── 20180314 第163期 │ └── README.md ├── 20180315 第164期 │ └── README.md ├── 20180316 第165期 │ └── README.md ├── 20180317 第166期 │ └── README.md ├── 20180318 第167期 │ └── README.md ├── 20180319 第168期 │ ├── README.md │ └── the-age-of-personal-assistants-more-machine-learning-less-hand-crafting.md ├── 20180320 第169期 │ ├── README.md │ └── neuronuggets-age-and-gender-estimation.md ├── 20180321 第170期 │ └── README.md ├── 20180322 第171期 │ └── README.md ├── 20180323 第172期 │ ├── README.md │ └── collecting-weather-data-to-boost-data-science-models-with-selenium.md ├── 20180324 第173期 │ └── README.md ├── 20180325 第174期 │ └── README.md ├── 20180326 第175期 │ ├── .md │ ├── README.md │ └── transfer-learning-vs-multitask-ibrahim.md ├── 20180327 第176期 │ └── README.md ├── 20180328 第177期 │ ├── README.md │ └── things-we-wish-we-had-known-before-we-started-our-first-machine-learning-project.md ├── 20180329 第178期 │ └── README.md ├── 20180330 第179期 │ └── README.md └── 20180331 第180期 │ ├── README.md │ └── introducing-tensorflow-js-machine-learning-in-javascript.md ├── 2018年4月 ├── 20180401 第181期 │ └── README.md ├── 20180402 第182期 │ ├── 50-must-read-books-for-machine-learning.md │ ├── README.md │ └── introducing-tensorflow-hub-a-library-for-reusable-machine-learning-modules-in-tensorflow.md ├── 20180403 第183期 │ └── README.md ├── 20180404 第184期 │ ├── README.md │ ├── highlights-from-tensorflow-developer-summit-2018.md │ ├── how-to-easily-detect-objects-with-deep-learning-on-raspberrypi.md │ └── instance-embedding-instance-segmentation-without-proposals.md ├── 20180405 第185期 │ ├── README.md │ └── from-zero-to-research-an-introduction-to-meta-learning.md ├── 20180406 第186期 │ ├── README.md │ └── dcnet-denoising-dna-sequence-with-a-lstm-rnn-and-pytorch.md ├── 20180407 第187期 │ └── README.md ├── 20180408 第188期 │ └── README.md ├── 20180409 第189期 │ ├── README.md │ └── toward-a-practical-perceptual-video-quality-metric.md ├── 20180410 第190期 │ ├── README.md │ └── maximum-likelihood-maximum-a-priori-and-bayesian-parameter-estimation.md ├── 20180411 第191期 │ └── README.md ├── 20180412 第192期 │ └── README.md ├── 20180413 第193期 │ ├── README.md │ └── sentiment-classification-from-keras-to-the-browser.md ├── 20180414 第194期 │ └── README.md ├── 20180415 第195期 │ └── README.md ├── 20180416 第196期 │ ├── README.md │ └── open-machine-learning-course-topic-9-time-series-analysis-in-python.md ├── 20180417 第197期 │ └── README.md ├── 20180418 第198期 │ └── README.md ├── 20180419 第199期 │ ├── README.md │ └── how-to-build-your-own-world-model-using-python-and-keras.md ├── 20180420 第200期 │ └── README.md ├── 20180421 第201期 │ └── README.md ├── 20180422 第202期 │ ├── README.md │ ├── artificial-intelligence-the-revolution-hasnt-happened-yet.md │ ├── open-machine-learning-course-topic-10-gradient-boosting.md │ └── what-do-we-learn-from-region-based-object-detectors-faster-r-cnn-r-fcn-fpn.md ├── 20180423 第203期 │ ├── README.md │ └── apple-ai-interview-questions-acing-the-ai-interview.md ├── 20180424 第204期 │ ├── 5-ways-of-looking-at-machine-learning-what-is-ml.md │ ├── README.md │ └── how-to-use-deep-learning-when-you-have-limited-data-part-2-data-augmentation.md ├── 20180425 第205期 │ └── README.md ├── 20180426 第206期 │ ├── README.md │ ├── adding-a-cutting-edge-deep-learning-training-technique-to-the-fast-ai-library.md │ └── predicting-the-price-of-wine-with-the-keras-functional-api-and-tensorflow.md ├── 20180427 第207期 │ └── README.md ├── 20180428 第208期 │ └── README.md ├── 20180429 第209期 │ ├── README.md │ └── transfer-learning-will-radically-change-machine-learning-for-engineers.md └── 20180430 第210期 │ ├── README.md │ └── sentiment-analysis-torchtext.md ├── 2018年5月 ├── 20180501 第211期 │ └── README.md ├── 20180502 第212期 │ └── README.md ├── 20180503 第213期 │ └── README.md ├── 20180504 第214期 │ ├── README.md │ └── the-two-steps-people-forget-to-take-when-applying-machine-learning.md ├── 20180505 第215期 │ ├── README.md │ └── colab-an-easy-way-to-learn-and-use-tensorflow.md ├── 20180506 第216期 │ └── README.md ├── 20180507 第217期 │ └── README.md ├── 20180508 第218期 │ ├── README.md │ └── deep-learning-and-time-to-predict-emojis.md ├── 20180509 第219期 │ ├── README.md │ └── real-time-human-pose-estimation-in-the-browser-with-tensorflow-js.md ├── 20180510 第220期 │ ├── README.md │ └── self-driving-cars-are-here.md ├── 20180511 第221期 │ ├── README.md │ └── multi-modal-methods-part-one.md ├── 20180512 第222期 │ └── README.md ├── 20180513 第223期 │ └── README.md ├── 20180514 第224期 │ └── README.md ├── 20180515 第225期 │ ├── README.md │ └── mxnet-for-pytorch-users-in-10-minutes.md ├── 20180516 第226期 │ └── README.md ├── 20180517 第227期 │ └── README.md ├── 20180518 第228期 │ └── README.md ├── 20180519 第229期 │ ├── README.md │ └── universal-word-sentence-embeddings.md ├── 20180520 第230期 │ └── README.md ├── 20180522 第231期 │ ├── README.md │ └── can-deep-networks-learn-invariants.md ├── 20180522 第232期 │ └── README.md ├── 20180523 第233期 │ ├── README.md │ └── mxboard-mxnet-data-visualization.md ├── 20180524 第234期 │ └── README.md ├── 20180525 第235期 │ └── README.md ├── 20180526 第236期 │ ├── README.md │ ├── comprehensive-guide-to-generative-adversarial-networks-and-wasserstein-gans.md │ └── exploring-deep-learning-and-neural-network-modeler-with-watson-studio.md ├── 20180527 第237期 │ ├── README.md │ └── a-simple-spell-checker-built-from-word-vectors.md ├── 20180528 第238期 │ └── README.md ├── 20180529 第239期 │ ├── README.md │ └── growing-up-with-ai-how-can-families-play-and-learn-with-their-new-smart-toys-and-companions.md ├── 20180530 第240期 │ └── README.md └── 20180531 第241期 │ └── README.md ├── 2018年6月 ├── 20180601 第242期 │ └── README.md ├── 20180602 第243期 │ └── README.md ├── 20180603 第244期 │ └── README.md ├── 20180605 第245期 │ └── README.md ├── 20180605 第246期 │ └── README.md ├── 20180606 第247期 │ └── README.md ├── 20180607 第248期 │ └── README.md ├── 20180608 第249期 │ └── README.md ├── 20180609 第250期 │ └── README.md ├── 20180610 第251期 │ └── README.md ├── 20180611 第252期 │ └── README.md ├── 20180612 第253期 │ └── README.md ├── 20180613 第254期 │ └── README.md ├── 20180614 第255期 │ ├── README.md │ └── the-end-of-video-coding.md ├── 20180615 第256期 │ └── README.md ├── 20180616 第257期 │ ├── README.md │ └── tensorflow-estimator-dataset-apis.md ├── 20180617 第258期 │ ├── README.md │ └── how-to-build-a-neural-network-framework-like-tensorflow-in-c-part-1.md ├── 20180618 第259期 │ └── README.md ├── 20180619 第260期 │ └── README.md ├── 20180620 第261期 │ ├── README.md │ ├── ncrf-deep-learning-framework-improves-cancer-metastasis-detection.md │ ├── neural-network-learning-part-1-feed-forward.md │ └── simple-website-text-scraping-with-go-and-aws-lambda.md ├── 20180622 第262期 │ ├── README.md │ └── multi-gpu-rosetta-stone.md ├── 20180622 第263期 │ ├── README.md │ └── why-you-should-start-using-in-place-activated-batchnorm.md ├── 20180623 第264期 │ └── README.md ├── 20180624 第265期 │ └── README.md ├── 20180626 第266期 │ └── README.md ├── 20180626 第267期 │ ├── README.md │ └── https-medium-com-100-shades-of-machine-learning-rediscovering-semantic-segmentation-part1.md ├── 20180628 第268期 │ └── README.md ├── 20180628 第269期 │ └── README.md ├── 20180629 第270期 │ ├── README.md │ └── tensorflow-cpus-and-gpus-configuration.md └── 20180630 第271期 │ └── README.md ├── 2018年7月 ├── 20180701 第272期 │ ├── README.md │ └── install-tensorflow-1-8-0-with-gpu-from-source-on-ubuntu-18-04-bionic-beaver.md ├── 20180702 第273期 │ └── README.md ├── 20180703 第274期 │ └── README.md ├── 20180704 第275期 │ └── README.md ├── 20180705 第276期 │ └── README.md ├── 20180707 第277期 │ ├── README.md │ └── an-introduction-to-biomedical-image-analysis-with-tensorflow-and-dltk.md ├── 20180707 第278期 │ ├── README.md │ └── cvpr-2018-impressions-and-15-interesting-papers.md ├── 20180708 第279期 │ └── README.md ├── 20180709 第280期 │ ├── README.md │ └── speed-up-tensorflow-inference-on-gpus-with-tensorrt.md ├── 20180710 第281期 │ └── README.md ├── 20180711 第282期 │ └── README.md ├── 20180713 第283期 │ └── README.md ├── 20180713 第284期 │ └── README.md ├── 20180714 第285期 │ └── README.md ├── 20180716 第286期 │ └── README.md ├── 20180716 第287期 │ └── README.md ├── 20180717 第288期 │ └── README.md ├── 20180718 第289期 │ └── README.md ├── 20180719 第290期 │ ├── README.md │ └── introducing-prose-v2-0-0-bringing-nlp-to-go.md ├── 20180721 第291期 │ ├── 100-times-faster-natural-language-processing-in-python.md │ ├── README.md │ └── move-mirror-an-ai-experiment-with-pose-estimation-in-the-browser-using-tensorflow-js.md ├── 20180722 第292期 │ ├── README.md │ ├── icml-2018-notes.md │ └── using-deep-learning-to-automatically-rank-millions-of-hotel-images.md ├── 20180722 第293期 │ └── README.md └── 20180723 第294期 │ └── README.md ├── README.md └── trans-docx ├── translated-.md.docx ├── translated-1.md.docx ├── translated-10 tips on using Jupyter Notebook.md.docx ├── translated-10-fearless-predictions-for-deep-learning-in-2018.md.docx ├── translated-100-times-faster-natural-language-processing-in-python.md.docx ├── translated-11行python实现神经网络.md.docx ├── translated-12-useful-things-to-know-about-machine-learning.md.docx ├── translated-13-deep-learning-frameworks-for-natural-language-processing-in-python.md.docx ├── translated-5-ways-of-looking-at-machine-learning-what-is-ml.md.docx ├── translated-50-must-read-books-for-machine-learning.md.docx ├── translated-8 Essential Tips for People starting a Career in Data Science.md.docx ├── translated-8-deep-learning-best-practices-i-learned-about-in-2017.md.docx ├── translated-Advice For New and Junior Data Scientists.md.docx ├── translated-AlphaGo与能力增强.md.docx ├── translated-Approaching (Almost) Any Machine Learning Problem Abhishek Thakur.md.docx ├── translated-Batch normalization in Neural Networks.md.docx ├── translated-Bayesian Learning for Statistical Classification.md.docx ├── translated-Best Practices for Document Classification with Deep Learning.md.docx ├── translated-Fundamentals of Deep Learning – Activation Functions and When to Use Them.md.docx ├── translated-Google Says Machine Learning Chips Make AI Faster and More Efficient.md.docx ├── translated-Hey Siri An On-device DNN-powered Voice Trigger for Apple’s Personal Assistant.md.docx ├── translated-How to Prepare Movie Review Data for Sentiment Analysis.md.docx ├── translated-Instacart Market Basket Analysis, Winner's Interview 2nd place, Kazuki Onodera.md.docx ├── translated-Intel® Nervana™ Neural Network Processors (NNP) Redefine AI Silicon.md.docx ├── translated-Natural Stupidity is more Dangerous than Artificial Intelligence.md.docx ├── translated-Planet Understanding the Amazon from Space, 1st Place Winner's Interview.md.docx ├── translated-Speech Recognition Is Not Solved.md.docx ├── translated-Subword-level embeddings.md.docx ├── translated-The difference between Statistical Modeling and Machine Learning, as I see it.md.docx ├── translated-Why AlphaGo Zero is a Quantum Leap Forward in Deep Learning.md.docx ├── translated-Word Tensors.md.docx ├── translated-Word embeddings in 2017 Trends and future directions.md.docx ├── translated-a-deeper-understanding-of-nnets-part-3-lstm-and-gru.md.docx ├── translated-a-genetic-algorithm-scheduled-our-developer-conference-heres-what-i-learnt.md.docx ├── translated-a-simple-spell-checker-built-from-word-vectors.md.docx ├── translated-adding-a-cutting-edge-deep-learning-training-technique-to-the-fast-ai-library.md.docx ├── translated-an-introduction-to-biomedical-image-analysis-with-tensorflow-and-dltk.md.docx ├── translated-apple-ai-interview-questions-acing-the-ai-interview.md.docx ├── translated-artificial-intelligence-the-revolution-hasnt-happened-yet.md.docx ├── translated-artwork-personalization.md.docx ├── translated-asking-the-right-questions-about-ai.md.docx ├── translated-ava-the-art-and-science-of-image-discovery-at-netflix.md.docx ├── translated-benchmarking-tensorflow-performance-on-next-generation-gpus.md.docx ├── translated-can-deep-networks-learn-invariants.md.docx ├── translated-colab-an-easy-way-to-learn-and-use-tensorflow.md.docx ├── translated-collecting-weather-data-to-boost-data-science-models-with-selenium.md.docx ├── translated-comprehensive-guide-to-generative-adversarial-networks-and-wasserstein-gans.md.docx ├── translated-cvpr-2018-impressions-and-15-interesting-papers.md.docx ├── translated-dcnet-denoising-dna-sequence-with-a-lstm-rnn-and-pytorch.md.docx ├── translated-deep-learning-and-time-to-predict-emojis.md.docx ├── translated-ethics-in-machine-learning.md.docx ├── translated-exploring-deep-learning-and-neural-network-modeler-with-watson-studio.md.docx ├── translated-first-time-with-kaggle-a-convnet-to-classify-toxic-comments-with-keras.md.docx ├── translated-from-research-to-practice.md.docx ├── translated-from-zero-to-research-an-introduction-to-meta-learning.md.docx ├── translated-getting-the-most-of-xgboost-and-lightgbm-speed-compiler-cpu-pinning.md.docx ├── translated-growing-up-with-ai-how-can-families-play-and-learn-with-their-new-smart-toys-and-companions.md.docx ├── translated-highlights-from-tensorflow-developer-summit-2018.md.docx ├── translated-how-to-build-a-neural-network-framework-like-tensorflow-in-c-part-1.md.docx ├── translated-how-to-build-your-own-alphazero-ai-using-python-and-keras.md.docx ├── translated-how-to-build-your-own-world-model-using-python-and-keras.md.docx ├── translated-how-to-create-a-machine-learning-framework-from-scratch-in-491-steps.md.docx ├── translated-how-to-easily-detect-objects-with-deep-learning-on-raspberrypi.md.docx ├── translated-how-to-use-deep-learning-when-you-have-limited-data-part-2-data-augmentation.md.docx ├── translated-how-we-grew-from-0-to-4-million-women-on-our-fashion-app-with-a-vertical-machine-learning-approach.md.docx ├── translated-https-medium-com-100-shades-of-machine-learning-rediscovering-semantic-segmentation-part1.md.docx ├── translated-icml-2018-notes.md.docx ├── translated-install-tensorflow-1-8-0-with-gpu-from-source-on-ubuntu-18-04-bionic-beaver.md.docx ├── translated-instance-embedding-instance-segmentation-without-proposals.md.docx ├── translated-introducing-prose-v2-0-0-bringing-nlp-to-go.md.docx ├── translated-introducing-tensorflow-hub-a-library-for-reusable-machine-learning-modules-in-tensorflow.md.docx ├── translated-introducing-tensorflow-js-machine-learning-in-javascript.md.docx ├── translated-its-a-no-brainer-deep-learning-for-brain-mr-images.md.docx ├── translated-logistic-regression-vs-deep-neural-networks-david.md.docx ├── translated-maximum-likelihood-maximum-a-priori-and-bayesian-parameter-estimation.md.docx ├── translated-meltdown.md.docx ├── translated-mltest-automatically-test-neural-network-models-in-one-function-call.md.docx ├── translated-move-mirror-an-ai-experiment-with-pose-estimation-in-the-browser-using-tensorflow-js.md.docx ├── translated-multi-gpu-rosetta-stone.md.docx ├── translated-multi-modal-methods-part-one.md.docx ├── translated-mxboard-mxnet-data-visualization.md.docx ├── translated-mxnet-for-pytorch-users-in-10-minutes.md.docx ├── translated-ncrf-deep-learning-framework-improves-cancer-metastasis-detection.md.docx ├── translated-neural-network-learning-part-1-feed-forward.md.docx ├── translated-neuronuggets-age-and-gender-estimation.md.docx ├── translated-nips-2017-policy-field-notes.md.docx ├── translated-only-numpy-why-i-do-manual-back-propagation-implementing-multi-channel-layer-convolution-neural.md.docx ├── translated-open-machine-learning-course-topic-10-gradient-boosting.md.docx ├── translated-open-machine-learning-course-topic-3-classification-decision-trees-and-k-nearest-neighbors.md.docx ├── translated-open-machine-learning-course-topic-9-time-series-analysis-in-python.md.docx ├── translated-otodscinos-the-root-cause-of-slow-neural-net-training.md.docx ├── translated-predicting-the-price-of-wine-with-the-keras-functional-api-and-tensorflow.md.docx ├── translated-real-time-human-pose-estimation-in-the-browser-with-tensorflow-js.md.docx ├── translated-self-driving-cars-are-here.md.docx ├── translated-sentiment-analysis-torchtext.md.docx ├── translated-sentiment-classification-from-keras-to-the-browser.md.docx ├── translated-simple-website-text-scraping-with-go-and-aws-lambda.md.docx ├── translated-snapchats-filters-how-computer-vision-recognizes-your-face.md.docx ├── translated-speed-up-tensorflow-inference-on-gpus-with-tensorrt.md.docx ├── translated-super-simple-end-to-end-test-of-keras-tensorflow-and-coreml.md.docx ├── translated-tensorflow-cpus-and-gpus-configuration.md.docx ├── translated-tensorflow-estimator-dataset-apis.md.docx ├── translated-the-3-tricks-that-made-alphago-zero-work.md.docx ├── translated-the-8-neural-network-architectures-machine-learning-researchers-need-to-learn.md.docx ├── translated-the-age-of-personal-assistants-more-machine-learning-less-hand-crafting.md.docx ├── translated-the-end-of-video-coding.md.docx ├── translated-the-two-steps-people-forget-to-take-when-applying-machine-learning.md.docx ├── translated-things-we-wish-we-had-known-before-we-started-our-first-machine-learning-project.md.docx ├── translated-toward-a-practical-perceptual-video-quality-metric.md.docx ├── translated-transfer-learning-vs-multitask-ibrahim.md.docx ├── translated-transfer-learning-will-radically-change-machine-learning-for-engineers.md.docx ├── translated-universal-word-sentence-embeddings.md.docx ├── translated-using-deep-learning-to-automatically-rank-millions-of-hotel-images.md.docx ├── translated-what-do-we-learn-from-region-based-object-detectors-faster-r-cnn-r-fcn-fpn.md.docx ├── translated-what-was-ai-in-2017.md.docx ├── translated-why-you-should-start-using-in-place-activated-batchnorm.md.docx ├── translated-使用YOLO提高实时目标检测能力.md.docx └── translated-如何使用C++写出一个TensorFlow.md.docx /2017年10月/20170925 第1期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---------------------------------------------------- | 3 | | [Deep Learning for Object Detection: A Comprehensive Review](https://towardsdatascience.com/deep-learning-for-object-detection-a-comprehensive-review-73930816d8d9?from=hackcv&hmsr=hackcv.com) | 介绍了目标检测模型的发展，包括R-CNN及其改进版本和SSD | 4 | | [Artificial intelligence pioneer says we need to start over](https://www.axios.com/artificial-intelligence-pioneer-says-we-need-to-start-over-1513305524-f619efbd-9db0-4947-a9b2-7a4c310a28fe.html) | Hinton提出需要发明新的方法来解决非监督学习的优化问题 | 5 | | [Search for the fastest Deep Learning Framework supported by Keras](https://www.axios.com/artificial-intelligence-pioneer-says-we-need-to-start-over-1513305524-f619efbd-9db0-4947-a9b2-7a4c310a28fe.html) | 比较开源的几大深度学习框架 | 6 | 7 | -------------------------------------------------------------------------------- /2017年10月/20170925 第1期/人工智能先驱说我们需要重新开始.md: -------------------------------------------------------------------------------- 1 | # 人工智能先驱说我们需要重新开始 2 | 3 | 原文链接：[Artificial intelligence pioneer says we need to start over](https://www.axios.com/artificial-intelligence-pioneer-says-we-need-to-start-over-1513305524-f619efbd-9db0-4947-a9b2-7a4c310a28fe.html) 4 | 5 | ![1](https://images.axios.com/voLHbhh_d9Erorx__kM7N9Jblug=/1920x1080/smart/2017/12/15/1513305524601.png) 6 | 7 | 杰弗里·辛顿怀疑人工智能当前的主力。 （Johnny Guatto /多伦多大学） 8 | 9 | 1986年，杰弗里·辛顿（Geoffrey Hinton）合著了一篇论文，三十年后，这篇论文对人工智能的爆发至关重要。但是Hinton说他的突破方法应该被摒弃，并且寻找了人工智能的新途径。 10 | 11 | 周三，在多伦多举行的人工智能会议期间与Axios谈话时，多伦多大学名誉教授，谷歌研究员希顿表示，他现在对[反向传播](https://www.iro.umontreal.ca/~vincentp/ift3395/lectures/backprop_old.pdf)“非常怀疑”，这是大多数人的主要方法。我们今天在人工智能领域取得的进步，包括整理照片和与[Siri交谈的能力](https://www.wired.com/2016/08/an-exclusive-look-at-how-ai-and-machine-learning-work-at-apple/)。 “我的看法是扔掉一切，重新开始，”他说。 12 | 13 | 14 | **核心内容**：会议上的其他科学家表示，反向传播在人工智能的未来仍然具有核心作用。但是，辛顿表示，要大力推进，可能必须要发明全新的方法。 “马克斯普朗克说，'科学推进一场葬礼。'未来由一些对我所说的一切都深表怀疑的研究生来决定。“ 15 | 16 | **工作原理**：在反向传播中，标签或“权重”用于表示类似大脑的神经层内的图像或声音。然后逐层调整和重新调整权重，直到网络能够以尽可能少的错误执行一个智能功能。 17 | 18 | 但是，Hinton建议，为了达到神经网络能够自己变得聪明的地方，即所谓的“无监督学习”，“我怀疑这意味着要摆脱反向传播。” 19 | 20 | “我不认为这是大脑的运作方式，”他说。 “显然我们不需要标记所有数据。” -------------------------------------------------------------------------------- /2017年10月/20171004 第2期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | :----------------------------------------------------------: | :--------------------------------------------: | 3 | | [Should Deep Learning use Complex Numbers?](https://medium.com/intuitionmachine/should-deep-learning-use-complex-numbers-edbd3aac3fb8?from=hackcv&hmsr=hackcv.com) | 介绍了一些论文关于深度学习是否使用复数方面的。 | 4 | | [How I started with learning AI in the last 2 months](https://hackernoon.com/how-i-started-with-learning-ai-in-the-last-2-months-251d19b23597?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 一个仅会Js的开发者走向AI学习的道路的故事。 | 5 | | [An Intuitive Explanation of Convolutional Neural Networks](https://ujjwalkarn.me/2016/08/11/intuitive-explanation-convnets/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 对于卷积神经网络的直观介绍。 | 6 | 7 | 8 | -------------------------------------------------------------------------------- /2017年10月/20171004 第2期/深度学习是否应使用复数？.md: -------------------------------------------------------------------------------- 1 | # 深度学习是否应该使用复数? 2 | 3 | 原文链接：[Should Deep Learning use Complex Numbers?](https://medium.com/intuitionmachine/should-deep-learning-use-complex-numbers-edbd3aac3fb8?from=hackcv&hmsr=hackcv.com) 4 | 5 | 大家不认为深度学习只能使用实数这很奇怪嘛？或许，深度学习使用复数才是更加奇怪的事情吧（注意：复数是有虚部的）。一个有价值的论点是：我们大脑在计算中不会使用复数。然而你也可以这样争论：大脑也不用矩阵运算或者链式法则微分啊。此外，人工神经网络（ANN）具有实际神经元的模型。长期以来，我们用实分析代替了生物合理性。深度学习的研究者发现线性代数和基本微积分足以展示开创性的结果，他们满足于此。 6 | 7 | 然而，为什么我们要止步于实分析呢？我们已经用了这么久线性代数和微分方程，那我们也可以将这一切都推倒，用复分析建立新的一套。或许更加奇妙的复分析会让我们找到更强大的方法。毕竟它对量子力学奏效，那么它也有可能在深度学习领域有所作为。此外，深度学习和量子力学都与信息处理有关，二者可能是同一件事情。 8 | 9 | 由于论据的原因，我们暂且不考虑生物合理性。这是一个很古老的观点，可以追溯到 1957 年 Frank Rosenblatt 第一次提出人工神经网络的时候。那么问题出现了，复数可以提供哪些实数不能提供的东西呢？ 10 | 11 | 在过去几年里，曾经出现过一些探索在深度学习中使用复数的文章。奇怪的是，它们中的大部分都没有被同行评议的期刊接受。因为深度学习的传统观念在该领域已经很流行了。但是，我们还是要讨论一些有趣的论文。 12 | 13 | DeepMind 有一篇论文《Associative Long Short-Term Memory》，文中探讨了使用复数值形成联想记忆神经网络。该系统被用来增强 LSTM （长短期记忆网络）的记忆。论文的结论是使用复数的网络可获取更大的记忆容量。根据数学原理，与只使用实数的情况相比，使用复数需要的矩阵更小。如下图所示，使用复数的神经网络在内存开销上与传统 LSTM 有显明显不同。 14 | 15 | Yoshua Bengio 和他的团队在Montreal探索了关于使用复数的另一方面。在一篇标题为“酉进化递归神经网络”（Martin Arjovsky,Amar Shah,Yoshua Bengio创造)的论文中，调查者探索了酉阵。他们认为如果矩阵的特征值接近1的话，那么在减少梯度的不扽或许会带来好处。在研究中，他们研究用复数作为RNN网络的权重。得到了这样的结论： 16 | 17 | 事实证明我们的uRNN能更好的通过长序列传递梯度信息，并且不会有像LSTM一样多的饱和隐藏状态。 18 | 19 | 他们做了多次实验去比较用复数的网络和传统 的RNN网络的性能：使用复数的网络明显比传统的有着更好、更稳定的表现。 20 | 21 | FAIL和EPEL的团队有一篇类似的论文《Kronecker Recurrent Units》，他们在论文里在复制任务中用酉阵的可行性。他们展示了一种能够大幅减少所需参数的矩阵分解方法。论文中描述了他们使用复数的动机。 22 | 23 | 因为实空间的行列式是连续函数，所以实空间的酉集是不连贯的。因而，使用标准的连续优化程序不能在实值网络上跨越全酉集。相反，酉集在复空间中是连接在一起的，因为它的行列式是复空间中单位圆上的点，所以使用复数的话就不会出现这个问题。 24 | 25 | 这篇论文的精华之一就是下面这个富有建设性的想法： 26 | 27 | 状态应当保持高维度，以此去使用高容量的网络将输入信息编码成内部状态、提取预测值。但 recurrent dynamic （递归动态） 可使用低容量模型实现。 28 | 29 | 目前，这些方法已经探索了在 RNN（循环神经网络） 上对复数值的使用。MILA（蒙特利尔学习算法研究所）最近的一篇论文《Deep Complex Networks》（Chiheb Trabelsi 等人）进一步探索了这些方法在卷积神经网络上的使用。论文作者在计算机视觉任务上测试了他们的网络，结果很有竞争力。卷积神经网络的创造者yann LeCun也有一篇相关的论文叫《A mathematical motivation for complex-valued convolutional networks》，这篇文章论述了使用复数的合理性。 30 | 31 | 最后，我们必须说一下复数在 GAN （生成式对抗网络） 中的使用。毕竟 GAN 可以说是最热的话题了。论文《Numerics of GANs》探讨了 GAN 中麻烦的收敛性能。他们研究了带有复数值的雅克比矩阵的特点，并使用它创建解决 GAN 均衡问题的最先进方法。 32 | 33 | 在去年的一篇博客中，我介绍了全息原理和深度学习的关系。文章中的方法探索了张量网络和深度学习架构网络之间的相似性。量子力学可以理解为是使用了一种更加通用的概率形式。复数的使用提供了常规概率无法提供的额外能力。具体来说就是叠加和干扰的能力。为了实现全息，在处理过程中使用复数会比较好。 34 | 35 | 在机器和深度学习空间中大量的数学分析倾向于使用贝叶斯思想作为观点。事实上，大多数从业者都认为它是贝叶斯的，但实际上它来自与统计学机制（除去名字，这里没有统计学的那些繁文缛节）。 36 | 37 | 但如果量子力学是广义的概率，那如果我们使用 QM 启发的方法替代会发生什么呢？一些论文试图研究这一方向，结果值得注意。在去年的一篇论文《Quantum Clustering and Gaussian Mixtures》中，作者探索了无监督硬聚类的使用情况。文章是这样说的： 38 | 39 | 因此，我们观察到了量子类干扰现象并不在高斯混合模型中出现。我们展示了量子方法在所有方面上都优于高斯混合方法。 40 | 41 | 下图为两者的对比： 42 | 43 | 噪声发生了什么？ 44 | 45 | 那么我们要想想，为什么在有了 20 世纪的量子概率理论后还要拘泥于 18 世纪的贝叶斯理论呢？（注意：令人震惊的是统计学家的货物崇拜自十八实际起就开始了。） 46 | 47 | 或许复数没有被经常使用的原因是研究者对它不够熟悉。在优化研究社区中，数学传统并没有涉及到复数。然而物理学家却一直在使用复数。那些虚部在量子力学中始终是存在的。这并不奇怪，这就是现实。我们仍然不太理解为何这些深度学习系统会如此有用。所以探索其他的表示可能会带来出乎意料的突破。 48 | 49 | 想法的替代可能会带来一些意想不到的图片。这就是我们的现在，偶然发现AGI(通用人工智能)突破的队伍获得胜利。 50 | 51 | 在不久的将来，这个局面可能会变化。最先进的结构可能会普遍使用复数，到那时候不使用复数反倒变得奇怪了。我想到了那个时候，18世纪的贝叶斯理论会完全过时。 52 | 53 | -------------------------------------------------------------------------------- /2017年10月/20171005 第3期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | :----------------------------------------------------------: | :-------------------------: | 3 | | [神经网络的入门](https://github.com/bhimmetoglu/talks-and-lectures/blob/master/MachineLearning/mnist/mnist_blog.ipynb) | 神经网络的介绍及代码实现。 | 4 | | [git git git git git](http://caiustheory.com/git-git-git-git-git/) | 一个关于git的有趣的小方法。 | 5 | 6 | -------------------------------------------------------------------------------- /2017年10月/20171005 第3期/git.md: -------------------------------------------------------------------------------- 1 | ## git git git git git 2 | 3 | 原文链接：[git git git git git](http://caiustheory.com/git-git-git-git-git/) 4 | 5 | 你是否曾在你的终端不小心进入太多的git，想找到一个有效的解决方法？我经常写了一个git然后离开一会，再回来写时又输入完整的git status 想查看它的状态。这会导致一个可爱的（烦人的）错误： 6 | 7 | ``` 8 | $ git git status 9 | git: 'git' is not a git command. See 'git --help'. 10 | ``` 11 | 12 | 什么是git。 13 | 14 | 我的初始想法是在我的$PATH中覆盖git的二进制，以此除去会导致报错的那些多余的git，这样的话我们最后运行的只是git status部分。一个简单的方法是使用git 的配置中的alias.*功能去扩展git使其变为一个shell 命令。 15 | 16 | ``` 17 | git config --global alias.git '!exec git' 18 | ``` 19 | 20 | 将下列git配置加入到你的 .gitconfig 文件中 21 | 22 | ``` 23 | [alias] 24 | git = !exec git 25 | ``` 26 | 27 | 然后你就可以git git 你想进行的内容。 28 | 29 | ``` 30 | $ git sha 31 | cc9c642663c0b63fba3964297c13ce9b61209313 32 | 33 | $ git git sha 34 | cc9c642663c0b63fba3964297c13ce9b61209313 35 | 36 | $ git git git git git git git git git git git git git git git git git git git git git git git git git git sha 37 | cc9c642663c0b63fba3964297c13ce9b61209313 38 | ``` 39 | 40 | (git sha 是 git rev-parse HEAD 的一个别名。) 41 | 42 | 看看我的 [~/.gitconfig][https://github.com/caius/zshrc/blob/master/dotfiles/gitconfig] 文件里还有什么别的Git 的别名，并嘲笑我那里面包含的所有拼写错误。（是的，git提供了自动改错如果你设定了的话，但是我习惯于让那些拼写错误的运行！） 43 | 44 | 现在去做有用的事情。 45 | 46 | -------------------------------------------------------------------------------- /2017年10月/20171006 第4期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | -------------------------------------------------- | 3 | | [Go: Ten years and climbing](https://commandcenter.blogspot.jp/2017/09/go-ten-years-and-climbing.html?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | Go语言十年回顾 | 4 | | [Why did we choose Rust over Golang or C/C++ to develop TiKV?](https://pingcap.github.io/blog/2017/09/26/whyrust/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 介绍作者为什么使用Rust开发TiKV而不用其他语言的原因 | 5 | | [Designing a Deep Learning Project](http://www.erogol.com/designing-deep-learning-project/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 介绍如何设计深度学习项目 | 6 | 7 | -------------------------------------------------------------------------------- /2017年10月/20171006 第4期/为什么我们开发TiKV选择Rust而不用Golang或者CC++.md: -------------------------------------------------------------------------------- 1 | # 为什么我们开发TiKV选择Rust而不用Golang或者C/C++? 2 | 3 | 原文链接：[Why did we choose Rust over Golang or C/C++ to develop TiKV?](https://pingcap.github.io/blog/2017/09/26/whyrust/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) 4 | 5 | ## 什么是Rust 6 | 7 | [Rust](https://en.wikipedia.org/wiki/Rust_(programming_language)) 是由Mozilla Research赞助的系统编程语言，自2015年5月的1.0版本以来，它以6周的发布周期快速稳定地运行。 8 | 9 | 请参阅下面的列表来了解最吸引我们的一些特性: 10 | 11 | - Rust的设计原则类似C++的 [没有消耗的抽象](https://blog.rust-lang.org/2015/05/11/traits.html) and [RAII ](https://rustbyexample.com/scope/raii.html)。 12 | - 最少的运行时间和绑定高效的C使得Rust能和C和C++一样高效，因此因此非常适合高性能最重要的系统编程领域。 13 | - 编译期间，强大的类型系统和独特的生命周期管理有助于内存管理，确保内存和线程的安全并且在编译后会让程序运行的非常迅速。 14 | - Rust提供字符串的匹配和像函数式编程语言那样的类型推断，使得代码简单、优雅。 15 | - 宏和特性允许Rust高度抽象，在工程期间，特别是在设计到库的时候节省了相当多的样板。 16 | 17 | ## Rust系统 18 | 19 | 由于Cargo这个优秀个包管理工具，Rust有很多类型的库，比如，用于HTTP的Hyper，异步I/O的Tokio和mio，基本上涵盖了构建后端应用程序的所需要的所有库。 20 | 21 | 一般来说，Rust主要用于开发具有高性能服务器的应用程序。 22 | 23 | ## 使用Rust 24 | 25 | 作为一个新的编程语言，Rust是独特的，仅举几个使用Rust的项目： 26 | 27 | - Dropbox的后端分布式存储系统 28 | - [Servo](https://github.com/servo/servo), Firefox的新内核 29 | - [Redox](https://github.com/redox-os/redox), 新的操作系统 30 | - [TiKV](https://github.com/pingcap/tikv), [TiDB](https://github.com/pingcap/tidb)的存储层, 由 [PingCAP](https://pingcap.com/index)开发的分布式数据库 31 | 32 | 33 | 34 | TiKV 是一个键值对分布式数据库，它是TiDB项目的核心组件，是Google Spanner的开源实现。在这个博客，我将会说出为什么我们选择Rust从头开始构建这样一个大型分布式存储项目。 35 | 36 | 在过去的很长一段时间，C或者C++主要开发类似数据库这样的基础设施软件，Java和Golang存在GC抖动等问题，尤其是在读/写压力较高的情况下。一方面，Goroutine的轻量级线程和Golang迷人的功能以其运行时切换上下文的额外开销为代价，明显降低了开发并发程序的复杂性。 37 | 38 | TiKV起源于2015年底。我们的团队在不同语言选择中苦苦挣扎，如Pure Go，Go + Cgo，C ++ 11或Rust。 39 | 40 | - **Pure Go:** 我们核新团队对于Go语言有着丰富的经验，TiDB的SQL层是用Go开发的，Go的性能很高。然而，当设计到存储层的开发的时候，Pure Go是第一个排除的语言，原因是：我们决定使用RocksDB作为底层，用C ++编写，Go中现有的LSM-Tree实现（如goleveldb）没有RocksDB那样成熟。 41 | 42 | - **Cgo:**如果我们使用Go，我们就要用Cgo来搭桥，但是Cgo有自己的问题。2015年年底，如果在Go代码中调用Cgo而不是在与Goroutine相同的线程中调用Cgo，性能可能会受到很大影响。此外，数据库需要频繁调用底层存储库，即RocksDB，如果每次调用RocksDB函数时都需要额外的开销，那么效率非常低。当然，可以引入一些变通方法来扩大调用Cgo的吞吐量，比如，在一定时间内为Cgo进行批量打包，这将减少单个请求的消耗而且抵消Cgo的开销。但是，GC问题还没有完全解决，这样实施可能会非常困难。在存储层，我们希望内存的速度尽可能的高效。Hacky的解决方法，例如广泛使用syscall.Mmap或对象重用可能会损害代码的可读性。 43 | 44 | - **C++11:** C ++ 11应该完全没有问题。 RocksDB是使用C ++ 11开发的。但鉴于团队背景和我们想要做的事情，我们没有选择C ++ 11。以下为理由： 45 | 46 | 1. 核心团队的成员是有着丰富的C++项目经验的资深C++开发者，但是像悬挂指针，内存泄漏或数据竞争这样的大型项目中看似不可避免的问题让他们对这个想法感到不寒而栗。当然，如果有良好的指导，或者严格的代码审查和编码规则，这些问题的可能性可能会降低很多。 47 | 2. C++有大量且复杂的编程范式以及太多的技巧，它需要额外的成本来统一编码风格，特别是当有越来越多的新成员可能不熟悉C ++。经过多年使用GC语言后，很难花时间手动管理内存。 48 | 3. 缺乏包管理和CI工具。自动化工具对于大的项目来说非常重要，因为关系到开发效率和迭代的速度，更重要的是，C ++库远远不够，其中一些需要由我们自己创建。 49 | 50 | - **Rust:** Rust的1.0版本于2015年5月发布，具有一些迷人的功能： 51 | 52 | 1. 内存安全 53 | 2. LLVM赋予的高性能。运行时几乎与C ++没有什么不同。它还与C / C ++包有密切关系。 54 | 3. Cargo, 强大的包管理工具 55 | 4. 现代语法 56 | 5. 几乎一致的故障排除和性能调整体验。我们可以直接重用一些我们已经非常熟悉的工具，比如perf。 57 | 6. FFI（外部函数接口），直接调用RocksDB中的C API而不会丢失。 58 | 59 | 内存安全是第一个而且也是最重要的原因。如前文所提，对于C++的老手而言，内存还礼和数据竞争的问题似乎变得很容易。但我相信Rust正在做的最大的解决方案是在编译器中加入约束并从一开始就解决它。对于大型的项目，永远不要用人来保证质量，人非圣贤，孰能无过。尽管对于Rust来说可能有点困难，但是我认为i这是完全值得的。此外，Rust是一种非常现代的编程语言，具有非凡的类型系统，模式建模，强大的宏，特征等。一旦熟悉它，就可以大大提高效率，这可能与我们选择C ++计算调试时间的效率相同。根据我们的经验，对于Rust零经验的软件工程师需要大概1个月来进行开发。经验丰富的Rust工程师和Golang工程师之间的效率几乎相同。 60 | 61 | 总而言之，作为一种新兴的编程语言，Rust似乎对中国的大多数开发人员来说都是新手，但它已成为C / C ++最有前途的挑战者。Rust也被称为“最受喜爱”的技术在 [StackOverflow’s 2016 developer survey](http://techbeacon.com/highlights-stack-overflow-2016-developer-survey).。因此从长远来看，Rust将在内存安全性和性能最重要的场景中大放异彩。 62 | -------------------------------------------------------------------------------- /2017年10月/20171006 第4期/设计深度学习项目.md: -------------------------------------------------------------------------------- 1 | ### 设计深度学习项目 2 | 3 | 原文链接：[Designing a Deep Learning Project](http://www.erogol.com/designing-deep-learning-project/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) 4 | 5 | 有大量关于深度学习的在线或者离线的技术资源，人们每天写新的东西，发布新的论文。然而，很少见这些关于深度学习的资源能够教我们实质上关心的问题--构建一个深度学习的项目；从上至下，从提出问题到解决问题。人们知道华丽的技术，但是一旦他们需要去构建自己项目，即使是有经验的人也会在细节中迷失。 6 | 7 | Andrew Ng’s 的一个新的计划 [deeplearning.ai](https://www.deeplearning.ai/) 可以帮助我们. [deeplearning.ai](https://www.deeplearning.ai/) 是一个有关深度学习所有有必要的细节(很适合初学者)的课程集合，而且其中有一门课程叫作"构建机器学习项目",特别的是，用现实生活中的例子来直观的设计一个深入学习的解决方案进行教授。这不可能给你个通用的方案，但是这个课程可以建立一个好的基础来思考DL的项目。显然，在我有多年经验后，我依旧可以从Andrew Ng.那学到知识。请注意，我也从他著名的Coursera ML课程开始了我的ML生涯。Thank you Mr. Ng. 8 | 9 | 下面，我试图画一个表格来总结在这门课程中所提到的。不过要注意，这不是按照课程逐字逐句进行记录，可能还包含了部分我自己的一些衡量标准。最后，这个对我而言是一个很好的练习，也希望，对你有一定的参考意义。 10 | 11 | 如果图表没有什么意义，请去看视频，如果你有什么不喜欢的地方，请告诉我。 12 | 13 | [![img](http://www.erogol.com/wp-content/uploads/2017/08/1-IsiFypDuJot1OHf4kxcqHA-720x1024.png)](http://www.erogol.com/wp-content/uploads/2017/08/1-IsiFypDuJot1OHf4kxcqHA.png) -------------------------------------------------------------------------------- /2017年10月/20171007 第5期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---------------------------------------------------------- | 3 | | [一种基于视觉词汇的文本分类方法](https://www.jianshu.com/p/f774e273a883?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 详细介绍了一个图像分类的具体操作步骤 | 4 | | [The Brain vs Deep Learning Part I: Computational Complexity — Or Why the Singularity Is Nowhere Near](http://timdettmers.com/2015/07/27/brain-vs-deep-learning-singularity/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 从各个角度对比，详细介绍了深度学习何时能达到大脑的计算水平 | 5 | | [EFF: Stupid patents are dragging down AI and machine learning](https://arstechnica.com/tech-policy/2017/10/eff-stupid-patents-are-dragging-down-ai-and-machine-learning/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 某些机器学习应用到一些显而易见的问题上，是否也能申请专利？ | 6 | -------------------------------------------------------------------------------- /2017年10月/20171007 第5期/愚蠢的专利正在拖垮人工智能和机器学习.md: -------------------------------------------------------------------------------- 1 | # 愚蠢的专利正在拖垮人工智能和机器学习 2 | 3 | 原文链接：[EFF: Stupid patents are dragging down AI and machine learning](https://arstechnica.com/tech-policy/2017/10/eff-stupid-patents-are-dragging-down-ai-and-machine-learning/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) 4 | 5 | **“专利读起来就像人工智能教科书简介的目录。”** 6 | 7 | ![](https://cdn.arstechnica.net/wp-content/uploads/2017/09/artificial-intelligence-resized-800x400.png) 8 | 9 | 每个月，电子前沿基金会的专利律师都会将[眼光瞄准](https://arstechnica.com/tech-policy/2017/05/effs-stupid-patent-of-the-month-dispatch-a-taxi-on-a-computer/)他们认为会影响创新的一项专利。本月，他们正在研究增长最快的技术领域之一：机器学习和人工智能。 10 | 11 | EFF律师Daniel Nazer选出了属于旧金山食品技术公司Hampton Creek的一项人工智能专利。该公司以“just”这个品牌销售产品。美国专利No.[9,760,834](https://www.google.com/patents/US9760834)介绍了[该公司称之为](http://www.businesswire.com/news/home/20170914006314/en)的“机器学习发现平台（machine-learning enabled discovery platform）”以及发现新成分的方法。 12 | 13 | 14 | 15 | 专利权利要求是长期的，因此有许多的具体规定被触犯。 但EFF的Daniel Nazer表示，该专利“反映了一种令人担忧的趋势”，因为这份冗长的专利申请要求在一种特定的应用程序上进行机器学习。在起诉过程中，Hampton Creek认为应该允许其专利，部分原因是早期的技术将机器学习应用于“分析数据”，而不是蛋白质片段。 16 | 17 | 专利权的主张是站在有利的一边的，因此为了侵犯专利权，人们必须做各种各样的具体事情。但EFF的丹尼尔·纳泽尔(Daniel Nazer)表示，这项专利“反映了一种令人担忧的趋势”，因为这份冗长的专利申请相当于在一种特定的应用程序上进行机器学习。在起诉过程中，Hampton Creek辩称，它的专利应该得到批准，部分原因是，早期的技术将机器学习应用于“分析数据”，而不是蛋白质片段。 18 | 19 | 20 | 21 | 还有一些借鉴了著名机器学习算法的声明。 22 | 23 | “事实上，我们认为该专利的内容就像AI教科书简介的目录，“Nazer写道。他继续说： 24 | 25 | > 它涵盖了几乎所有你希望在AI介绍中学习的标准机器学习技术 - 包括线性和非线性回归，k-最近邻，聚类，支持向量机，主成分分析，基于lasso的特征选择或弹性网，高斯过程，甚至决策树 - 且应用于蛋白质这些可测量数据的具体示例。 当然，将这些技术应用于蛋白质可能是一项有价值且耗时的工作。 但这并不意味着它应该获得专利。 26 | 27 | Nazer承认Hampton Creek的专利并不像EFF愚蠢的专利系列中提到的其他一些专利那么糟糕，但值得指出的是，它可能为机器学习的创新带来严重的问题。 28 | 29 | 30 | 31 | 就像美国专利局(US Patent Office)过去在计算机上统计投票或统计卡路里等简单操作方面提出的专利问题一样，专利局似乎准备提供“以明显和预期的方式使用机器学习”的专利。 Nazer写道，像谷歌和微软这样的公司正在尝试申请“基本机器学习技术”的专利，并已经在某些方面获得了这些专利。 32 | 33 | 34 | 35 | Hampton Creek的一位发言人拒绝对EFF邮报发表评论。本月早些时候，就在专利发布后，该[公司](http://www.businesswire.com/news/home/20170914006314/en)发布了一份新闻稿，称该专利涵盖了公司的“独一无二的机器人技术、专有的植物数据库、人工智能和预测建模”，并将其整合到一个名为黑鸟(Blackbird)的系统中。 36 | -------------------------------------------------------------------------------- /2017年10月/20171008 第6期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | :----------------------------------------------------------: | :----------------------------------------------------------: | 3 | | [NEURAL NETWORK GENERATES FAKE STARTUPS. HILARITY ENSUES](https://www.topbots.com/recurrent-neural-network-generates-startup-names-hilarity-ensues/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 神经网络的一个有趣的应用。 | 4 | | [COMPUTER VISION IN IOS – OBJECT DETECTION](https://sriraghu.com/2017/07/12/computer-vision-in-ios-object-detection/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | ios系统实现目标识别的代码教程。 | 5 | | [Is AI Riding a One-Trick Pony?](https://www.technologyreview.com/s/608911/is-ai-riding-a-one-trick-pony/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | AI的发展相关：反向传播的重要性和局限性，AI下一个进步的关键是哪里？ | 6 | 7 | -------------------------------------------------------------------------------- /2017年10月/20171009 第7期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---- | 3 | | [5 Ways the Internet of Things Is Revolutionizing Healthcare](https://www.makeuseof.com/tag/internet-things-revolutionizing-healthcare/) | | 4 | | [Recurrent Neural Networks](https://freecontent.manning.com/recurrent-neural-networks/) | | -------------------------------------------------------------------------------- /2017年10月/20171010 第8期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | -------------------------------------------------------- | 3 | | [An Introduction to different Types of Convolutions in Deep Learning](https://towardsdatascience.com/types-of-convolutions-in-deep-learning-717013397f4d?from=hackcv&hmsr=hackcv.com) | 介绍深度学习中各种卷积 | 4 | | [ML Algorithms addendum: Passive Aggressive Algorithms](https://www.bonaccorso.eu/2017/10/06/ml-algorithms-addendum-passive-aggressive-algorithms/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 介绍了以一种名为被动攻击的机器学习算法，可用于分类和回归 | 5 | | [我需要知道：H.264](https://blog.piasy.com/2017/09/22/I-Need-Know-About-H264/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 6 | 7 | -------------------------------------------------------------------------------- /2017年10月/20171010 第8期/在深度学习中关于不同卷积的介绍.md: -------------------------------------------------------------------------------- 1 | # 在深度学习中关于不同卷积的介绍 2 | 3 | 原文链接：[An Introduction to different Types of Convolutions in Deep Learning](https://towardsdatascience.com/types-of-convolutions-in-deep-learning-717013397f4d?from=hackcv&hmsr=hackcv.com) 4 | 5 | 让我为你简单概括不同类型卷积和它们优点所在。为了简单起见。我将把重点仅放在2D卷积上。 6 | 7 | ## 卷积 8 | 9 | 首先，我们需要就定义卷积层的一些参数达成一致。 10 | 11 | 这里2D卷积使用内核大小为3，步长为1且需填充. 12 | 13 | - __内核大小__ ：内核大小定义了卷积的视野视图大小。2D的常见选择内核大小为3即3*3像素。 14 | 15 | - __步长__：步长定义了当遍历图像时每一步大小。虽然它的默认值通常为1，但我们也能使用步长为2来对与MaxPooling类似的图像进行下采样。 16 | 17 | - __填充__：填充定义如何处理样本的边界。（半）填充卷积将使空间输出维数保持等于其输入维数，然而如果内核大于1，则未填充卷积将裁掉一些边界。 18 | - __输入和输出通道__: 卷积层占用一定数量的输入通道（I）并计算特定数量的输出通道（O）。可以通过I *O* K 计算这一层所需的参数，其中K等于内核中的值的数量。 19 | 20 | ## 扩张的卷积 21 | 22 | (又名空洞卷积) 23 | 24 | 2D卷积所使用为3内核进行2D卷积，扩展率为2且无填充。 25 | 26 | 扩张的卷积为卷积层引入另一个参数，称为**扩张率**。这定义了内核中值之间的间距。扩散率为2的3x3内核与5x5内核具有相同的视野，而仅使用9个参数。想象一下，获取一个5x5内核并删除每一个第二列和一行。 27 | 28 | 这以相同的计算成本提供了更宽的视野。扩张卷积在实时分割领域中特别受欢迎。如果您需要广泛的视野并且无法承受多个卷积或更大的内核，请使用它们。 29 | 30 | ## 转置卷积 31 | 32 | （又称解卷积或分数跨度卷积） 33 | 34 | 有些来源使用名称deconvolution，这是不合适的，因为它不是解卷积。为了使事情更糟，确实存在解卷积，但它们在深度学习领域并不常见。实际的反卷积会使卷积过程恢复。想象一下，将图像输入到单个卷积层中。现在取出输出，将它扔进一个黑盒子里然后再出现原始图像。这个黑盒子进行反卷积。它是卷积层的数学逆。 35 | 36 | 转置卷积有点类似，因为它产生与假设的反卷积层相同的空间分辨率。但是，对值执行的实际数学运算是不同的。转置卷积层执行常规卷积，注意会恢复其空间变换。 37 | 38 | 2D卷积，没有填充，步幅为2，内核为3 39 | 40 | 此时你应该很困惑，让我们看一个具体的例子。将5×5的图像送入卷积层。步幅设置为2，填充停用，内核为3x3。这会导致出现2x2图像。 41 | 42 | 如果我们想要反转这个过程，我们需要逆数学运算，以便从我们输入的每个像素生成9个值。然后，我们以2的步幅遍历输出图像。这将是反卷积。、 43 | 44 | 2D转置卷积，没有填充，步幅为2，内核为3 45 | 46 | 转置卷积不会这样做。唯一的共同点是它保证输出也是5x5图像，同时仍然执行正常的卷积操作。为此，我们需要在输入上执行一些复杂的填充。 47 | 48 | 正如您现在可以想象的那样，此步骤不会从上面颠倒过程。至少不涉及数值。 49 | 50 | 它只是从之前重建空间分辨率并执行卷积。这可能不是数学逆，但对于编码器 - 解码器架构，它仍然非常有用。这样我们就可以将图像的升级与卷积相结合，而不是进行两个单独的处理。 51 | 52 | ## 可分离卷积 53 | 54 | 在可分离的卷积中，我们可以将内核操作分成多个步骤。让我们将卷积表示为**y = conv（x，k）**，其中**y**是输出图像，**x**是输入图像，**k**是内核。很简单吧。那么接下来，假设k可以通过以下公式计算：**k = k1.dot（k2）**。因为我们可以通过用k1和k2进行2个1D卷积来得到相同的结果，而不是用简单k进行2D卷积，所以这将使它成为可分离的卷积， 55 | 56 | Sobel X和Y滤镜 57 | 58 | 以Sobel内核为例，它通常用于图像处理。你可以通过乘以向量[1,0，-1]和[1,2,1] .T 得到相同的内核。在执行相同操作时，这将需要6个而不是9个参数。上面的例子显示了所谓的**空间可分卷积**，据我所知，它不用于深度学习。 59 | 60 | *编辑：实际上，通过堆叠1xN和Nx1内核层，可以创建与空间可分离卷积非常相似的东西。这最近在一个名为* [*EffNet*](https://arxiv.org/abs/1801.06434v1) *的架构中使用，**显示了有希望的结果。* 61 | 62 | 在神经网络中，我们通常使用称为**深度可分离卷积的**东西**。**这将执行空间卷积，同时保持通道分离，然后进行深度卷积。在我看来，通过一个例子可以最好地理解它。 63 | 64 | 假设我们在16个输入通道和32个输出通道上有一个3x3卷积层。详细情况是，32个3x3内核遍历16个通道中的每一个，产生512（16x32）个特征映射。接下来，我们通过添加它们来合并每个输入通道中的1个特征图。由于我们可以做32次，我们得到了我们想要的32个输出通道。 65 | 66 | 对于同一示例中的深度可分离卷积，我们遍历16个通道，每个通道有1个3x3内核，为我们提供16个特征映射。现在，在合并任何东西之前，我们遍历这16个特征映射，而每个特征映射有32个1x1卷积，然后才开始将它们加在一起。这导致656（16x3x3 + 16x32x1x1）参数与上面的4608（16x32x3x3）参数相反。 67 | 68 | 该示例是深度可分离卷积的特定实现，其中所谓的**深度乘数**为1.这是迄今为止这种层的最常见设置。 69 | 70 | 我们这样做是因为空间和深度信息可以解耦的假设。看一下Xception模型的表现，这个理论似乎有效。由于其有效使用参数，深度可分离卷积也用于移动设备。 71 | 72 | ### 有问题吗？ 73 | 74 | 这就结束了我们通过不同类型的卷积进行的小游览。我希望有助于对此事进行简要概述。如果您有任何剩余的问题，请发表评论，并查看[此](https://github.com/vdumoulin/conv_arithmetic) GitHub页面以获取更多卷积动画。 75 | 76 | -------------------------------------------------------------------------------- /2017年10月/20171011 第9期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---- | 3 | | [tensorflow](https://github.com/nicodjimenez/nicodjimenez.github.io/blob/master/_posts/2017-10-08-tensorflow.markdown) | | 4 | | [The limitations of deep learning](https://blog.keras.io/the-limitations-of-deep-learning.html) | | 5 | | [揭秘支付宝中的深度学习引擎：xNN](https://mp.weixin.qq.com/s/ZuEi32ZBSjruvtyUimBgxQ?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 6 | 7 | -------------------------------------------------------------------------------- /2017年10月/20171011 第9期/深度学习的局限性.md: -------------------------------------------------------------------------------- 1 | # 深度学习的局限性 2 | 原文链接：[The limitations of deep learning](https://blog.keras.io/the-limitations-of-deep-learning.html) 3 | 4 | 本章节改编自我一本书的第九章第2部分，[Deep Learning with Python](https://www.manning.com/books/deep-learning-with-python?a_aid=keras&a_bid=76564dff) (Manning Publications) 5 | 6 | [![Deep learning with Python](https://blog.keras.io/img/deep_learning_with_python_cover_thumbnail.png)](https://www.manning.com/books/deep-learning-with-python?a_aid=keras&a_bid=76564dff) 7 | 8 | 这是一系列两篇关于当前深度学习局限性和未来发展文章的一部分。 9 | 10 | 本文适合已经拥有重要深度学习经验的人（比如说那些已经阅读了本书一到八章节的人），我们假定你已经拥有许多前置知识。 11 | 12 | # 深度学习：几何视图 13 | 最令人惊讶的事情是深度学习如此简单。十年前，没有人期望我们能够通过使用简单的、用梯度下降训练的参数模型，在机器感知问题方面实现如此惊人的成果。现在，你只需要有足够大量的参数模型，并用足够多的示例进行梯度下降训练。就如Feynman曾经谈论过宇宙，“它并不复杂，它只是很多”。 14 | 15 | 在深度学习里，一切事物都是一个向量，即一切事物都是几何空间中的一个点。模型输入（它可以是文本、图片等等）和目标首先会被向量化，即转化成一些初始的输入向量空间和目标向量空间。当数据经过深度学习中的每个层面，都会做一次简单的几何变换。总之，模型的层链形成了一个非常复杂的几何变换，它们可以分解为一系列的简单几何变换。这个复杂的转换企图将输入空间映射到目标空间，一次一个点。这个转换被每个层面的权重参数化，这些权重基于模型当前表现的良好程度进行更新。这种几何变换的一个关键特性是它必须是可微分的，目的是让我们能够通过梯度下降训练来学习它的参数。直观地说，这意味着从输入到输出的几何变形必须是平滑而且连续的——这是一个重要的约束。 16 | 17 | 对输入数据应用复杂的几何变换的这整个过程可以被3D可视化，通过想象一个尝试将皱纸球解皱的人：皱纸球是模型开始时输入数据的歧管，在纸球上被人操作的每一个动作类似于被一个层面操作的一个简单几何变换，这整个解皱手势序列就是整个模型的复杂变换。深度学习模型是用于解开复杂的高维数据流形的数学机器。 18 | 19 | 这就是深度学习的神奇之处：把含义转换成向量、几何空间，然后学习复杂的几何变换，将一个空间映射到另一个空间。你需要的仅仅是足够高的维度，用于捕捉在原始数据中找到的关系的全部范围。 20 | 21 | 22 | # 深度学习的局限性 23 | 用这种简单的策略实现的应用空间几乎是无穷的，然而，更多的应用是现在深度学习技术不能实现的——就算是给出了大量的人工注释的数据。例如，假设您可以组装一个数十万甚至数百万个软件产品功能的英语描述的数据集，由产品经理编写，以及由团队工程师开发的相应源代码来满足这些需求。即使有了这些数据，您也不能通过简单地训练一个深度学习模型来读取产品描述并生成适当的代码库。这只是许多例子中的一个例子。一般来说，任何需要推理（如编程）、应用科学方法（如长期规划）或者类似算法的数据操纵的东西，对于深度学习模型来说都是遥不可及的，不管你向它扔了多少数据。即使学习具有深度学习网络的排序算法也是相当困难的。 24 | 25 | 这是因为深度学习模型只是通过一系列简单的、连续的几何变换将一个向量映射到另一个向量空间。假设存在从X到Y到可学习的连续变换，以及可用作训练数据的X:Y的密集采样，它所能做的就是将一个数据流形X映射到另一个流形Y。因此，即使深度学习模型可以理解为一种程序，但是，大多数程序不能用深度学习模型所表达——这是对于大多数任务而言，或者不存在相应的解决该任务的实际大小的深层神经网络，或者即使存在一种情况，它可能也不是可学习的，即相应的几何变换可能太复杂，或者可能没有合适的数据可以用来学习它。 26 | 27 | 通过堆叠更多的层和使用更多的训练数据来扩展当前的深度学习技术，只能从表面上缓解这些问题。它不能解决更基本的问题，即深层学习模型在表示内容方面非常有限，并且人们希望能够学习的大多数程序不能表示为数据流形的连续几何变形。 28 | 29 | # 机器学习拟人化的风险 30 | 31 | 当代人工智能的一个非常真实的风险是误解深度学习模型的作用，并且高估了它们的能力。人类头脑的一个基本特征是"头脑理论"，即我们倾向于投射关于我们周围事物的意图、信仰和知识。在岩石上画一张笑脸使它"快乐"在我们心中。例如，应用于深度学习，这意味着当我们能够稍微成功地训练一个模型来生成描述图片的字幕时，我们被引导相信该模型"理解"图片的内容及它所生成的字幕。然后，当训练数据中出现的图像稍有偏离导致模型开始生成完全荒谬的字幕时，我们就会感到非常惊讶。 32 | 33 | -------------------------------------------------------------------------------- /2017年10月/20171012 第10期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---- | 3 | | [A Research to Engineering Workflow](http://dustintran.com/blog/a-research-to-engineering-workflow?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 4 | | [Introduction to Information Theory and Why You Should Care](https://recast.ai/blog/introduction-information-theory-care/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 5 | | [A Gentle Introduction to the Bag-of-Words Model](https://machinelearningmastery.com/gentle-introduction-bag-words-model/) | | 6 | 7 | -------------------------------------------------------------------------------- /2017年10月/20171013 第11期/PyTorch与TensorFlow 一个月总结.md: -------------------------------------------------------------------------------- 1 | # PyTorch与TensorFlow: 一个月总结 2 | 3 | 原文链接：[PyTorch vs. TensorFlow: 1 month summary](https://towardsdatascience.com/pytorch-vs-tensorflow-1-month-summary-35d138590f9?from=hackcv&hmsr=hackcv.com) 4 | 5 | *在使用PyTorch一个月后，比较PyTorch与TensorFlow。* 6 | 7 | ![img](https://cdn-images-1.medium.com/max/800/1*OFhhTYPS42zjyabSVdBQmQ.jpeg) 8 | 9 | 我一直是TensorFlow用户，将其更好的用于深度学习工作。但是，当我加入[NVIDIA时](https://medium.com/@NvidiaAI)，我们决定改用PyTorch - 做一个测试一样。以下是我的经历。 10 | 11 | ### 安装 12 | 13 | 安装非常简单直接。PyTorch可以通过PIP安装，也可以编译源代码。PyTorch还提供Docker镜像，可用作您自己项目的基本镜像。 14 | 15 | PyTorch没有指定的CPU和GPU版本，就像TensorFlow一样。虽然这让安装过程更容易，但如果您想同时支持CPU和GPU使用，它会需要更多代码。 16 | 17 | 值得一提的是，PyTorch尚未提供正式的Windows发行版。Windows有非官方端口，但没有PyTorch的支持。 18 | 19 | ### 使用 20 | 21 | PyTorch提供了一个非常Pythonic的API。这与TensorFlow非常不同，TensorFlow是定义了所有的Tensors和Graph，然后在会话（session）中运行它。 22 | 23 | 在我看来，这会增加代码量但是增加了可读性。PyTorch图必须在继承自PyTorch `nn.Module`的类中定义。`forward()`运行Graph时会调用一个函数。使用这种“约定优于配置”方法，Graph的位置始终是已知的，并且在其余代码中并未定义变量。 24 | 25 | 这种“新”方法需要一些时间来习惯，但我认为如果您之前在深度学习之外运用过Python，那将非常直观。 26 | 27 | 基于一些评论，与TensorFlow相比，PyTorch在许多模型上也表现出更好的性能 28 | 29 | ### 文档 30 | 31 | 文档大部分都是完整的。我没有找不到的函数或模块的定义。与TensorFlow相反，所有函数都有一个页面，PyTorch每个模块都只使用一个页面。如果去Google寻找函数，反而会更加困难。 32 | 33 | ### 社区 34 | 35 | 显然，PyTorch的社区并不像TensorFlow那么大。然而，许多人喜欢在业余时间使用PyTorch，即使他们使用TensorFlow进行工作。我认为一旦PyTorch完成公测（Beta），这种情况就会发生变化。 36 | 37 | 目前，在PyTorch中找到精通它的人仍然有点困难。 38 | 39 | 但社区还是足够大的，官方论坛中的问题通常可以很快地收到回复，因此很多很棒网络模型的示例实现都使用了PyTorch。 40 | 41 | ### 工具和助手 42 | 43 | 即使PyTorch提供了大量的工具，也缺少一些非常有用的工具。缺少最有用的工具之一是TensorFlow的TensorBoard。这使得可视化（vizualization）有点困难。 44 | 45 | 还有一些非常常见的使用助手丢失。这需要自己编写比TensorFlow更多的代码。 46 | 47 | ### 结论 48 | 49 | PyTorch是TensorFlow的一个很棒的替代品。由于PyTorch仍然处于测试阶段，我希望对可用性，文档和性能进行一些更改和改进。 50 | 51 | PyTorch非常pythonic，使用起来很舒服。它有一个很好的社区和文档。它也被认为比TensorFlow快一点。 52 | 53 | 但是，与TensorFlow相比，社区仍然相当小，并且缺少一些有用的工具，如TensorBoard。 -------------------------------------------------------------------------------- /2017年10月/20171013 第11期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ----------------------------------------------------- | 3 | | [PyTorch vs. TensorFlow: 1 month summary](https://towardsdatascience.com/pytorch-vs-tensorflow-1-month-summary-35d138590f9?from=hackcv&hmsr=hackcv.com) | 作者介绍了他从TensorFlow转用PyTorch一个月后的对比总结 | 4 | | [Integrating convolutional neural networks into enterprise applications](https://www.oreilly.com/ideas/integrating-convolutional-neural-networks-into-enterprise-applications?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 介绍如何将卷积神经网络落地，考虑了各种复杂的情况 | 5 | | [10 Common Software Architectural Patterns in a nutshell](https://towardsdatascience.com/10-common-software-architectural-patterns-in-a-nutshell-a0b47a1e9013?from=hackcv&hmsr=hackcv.com) | 简单介绍了流行的10种软件架构模式 | 6 | 7 | -------------------------------------------------------------------------------- /2017年10月/20171014 第12期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---- | 3 | | [Visual Information Theory](http://colah.github.io/posts/2015-09-Visual-Information/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 入门信息论的详细介绍并配上大量图片辅助理解 | 4 | | [Pipeline Patterns in Go](https://medium.com/statuscode/pipeline-patterns-in-go-a37bb3a7e61d?from=hackcv&hmsr=hackcv.com) | 详细介绍了Go的管道模式并附上了代码 | 5 | | [High Time to Regulate Face Recognition A.I.](https://medium.com/intuitionmachine/high-time-to-begin-regulation-of-face-recognition-a-i-f4a92ee40165?from=hackcv&hmsr=hackcv.com) | 介绍了关于人脸识别应用的风险与好处，作者提出了风险大于好处的观点 | -------------------------------------------------------------------------------- /2017年10月/20171014 第12期/管理人脸识别AI的时间还很长.md: -------------------------------------------------------------------------------- 1 | # 管理人脸识别AI的时间还很长 2 | 3 | 原文链接：[High Time to Regulate Face Recognition A.I.](https://medium.com/intuitionmachine/high-time-to-begin-regulation-of-face-recognition-a-i-f4a92ee40165?from=hackcv&hmsr=hackcv.com) 4 | 5 | 我们已经到了一个转折点，现在正是我们开始调整人脸识别人工智能（AI）的对话的时候了。 6 | 7 | 在上一篇[文章中](https://medium.com/intuitionmachine/how-to-regulate-artificial-intelligence-10725701043b)，我探讨了一些关于如何规范人工智能的想法。我查看了其他领域的几个法规，并探讨了如何申请AI。反对人工智能监管最引人注目的论点是，对于许多人来说，确切地说需要监管的内容并不清楚。然而，最近几天，我注意到一种特定的AI算法需要认真考虑规则。 8 | 9 | [斯坦福大学的研究人员](https://osf.io/zn79k/)已经培训了一种深度学习系统来识别一个人的性取向。我不怀疑他们建立的系统的有效性。然而，许多人质疑研究人员的解释。但是，我非常关注滥用这种技术的可能性。 10 | 11 | 12 | 13 | ![img](https://cdn-images-1.medium.com/max/800/1*5VhzZSfzf-ETGjy7fx-abQ.jpeg) 14 | 15 | 来源: [Deep neural networks are more accurate than humans at detecting sexual orientation from facial images.](https://osf.io/zn79k/) 16 | 17 | 今年5月，中国研究人员开发了一种更加值得怀疑的面部分类系统，确定某人是否可能成为犯罪分子。 18 | 19 | ![img](https://cdn-images-1.medium.com/max/800/1*EEhfnjTH_nNHDwiXwwby8g.jpeg) 20 | 21 | 论文: [Automated Inference on Criminality using Face Images](https://arxiv.org/pdf/1611.04135.pdf) 22 | 23 | 我们现在迫切需要分析面部识别的好处大于风险。我将对此进行一次尝试，并最终提出风险大于收益的论点。 24 | 25 | 以下是面部识别的一些好处： 26 | 27 | **安全性**  - 面部识别现已内置于iPhone X中（请参阅：[面部识别码](https://www.wired.com/story/iphone-x-faceid-security/)），并用作解锁手机的身份验证机制。以前，iPhone（其他智能手机）使用一个指纹来解锁设备。面部识别还可用于监控访问家庭或办公室的人。 28 | 29 | **社交参与**  - Facebook可以自动标记照片中的面孔。此功能的价值在于增加被标记的人以及跟随其他人的人的参与度。Facebook增加了额外的安全功能，允许用户选择退出自动面部标记功能。 30 | 31 | **安全性** - 在车身监控设备中可以确定驾驶员是否暂时分心。该申请可用于任何需要受雇工人集中注意力的工作相关活动。 32 | 33 | **医疗保健**  - 为生病或残疾人士提供的监控设备可能有助于为他们的需求提供更具响应性的医疗服务。 34 | 35 | **娱乐**  - 虚拟现实或增强现实游戏可能会对参与者的情绪做出反应。 36 | 37 | **生产力**  - Google相册使用面部识别功能自动整理和搜索自己的个人照片。 38 | 39 | **执法**  - 面部识别可用于识别刑事调查中的嫌疑人。这也可用于跟踪和查找失踪人员。 40 | 41 | 以下是潜在风险： 42 | 43 | **法治国家**  - 面部识别允许各州追踪其公民的行动和行为。这允许各州强制遵守其议程。 44 | 45 | **生物识别**  - 面部识别可以识别一个人的性取向。在许多国家，一个人的性取向可以作为犯罪行为受到惩罚。这样的系统可以变得多么准确，这可能是有争议的。但是，如果政府对其准确性感到满意，那么人们就会因为他们是谁而不是他们所做的而受到惩罚。识别到一个人的传承或种族是另一个可能被滥用的领域。 46 | 47 | **行为操纵**  - 这涉及在不同背景下识别人的面部反应。识别厌恶或敬畏之类的反应可能向观察者表明一个人的内在个人观点。利用这些信息允许其他方（即营销，销售，政府，雇主等）执行可能与个人福祉相冲突的行为操纵技术。我们已经在社交网络中见证了这些行为技巧，以影响一个人的行为。 48 | 49 | **行为执法**  - 可以对工人进行监控，以评估他们对手头任务的持续关注。不认真工作将受到扣发工资的惩罚。 50 | 51 | 关于风险，我现在就在这里停下来。有很多风险在这里我故意没有提到。人们甚至可以进行推断在未来的情况，很容易想象出一种我们会后悔让机器识别我们的情景。 52 | 53 | 然而，为什么面部识别是一个如此危险的想法，这里的一般总体原则是什么？第一个问题是因为它是通过自动化执行的，所以它可以大规模地完成并且毫无感情地完成。这导致了我们与大型互联网公司相同的隐私问题。第二个问题也是隐私问题，即我们内心的想法和偏好应该保密。第三个问题是机器进行简易判断的可能性。对于人类来说，懒惰地依赖算法来执行判断而不是自己这样做是非常容易的。此外，有一个算法做出判断，在许多方面清洗一个人来清除任何后果的责任。当判断标准基于你的外表时，这种判断会更加糟糕。 54 | 55 | 照片复印机受到管制，无法复制和打印合法纸币。任何打印机中的软件都是硬连线的，用于检测货币中的特定签名，从而防止传真复制。类似地，可能需要要求任何新的深度学习芯片（具有足够的性能）使任何面部识别任务无效。或者，在没有必要的“安全芯片”的情况下，将深度学习设备与图像捕获系统连接起来应该是非法的。 56 | 57 | 较轻的监管形式应允许任何人选择退出面部识别。它可以通过使用一种用户佩戴的标记或代码（可能是额头上的条形码？）来实现。这类似于美国发现的自我识别种族形式。根据法律，任何人都应该被允许选择退出面部识别软件。替代方案将是一种新的时尚： 58 | 59 | ![img](https://cdn-images-1.medium.com/max/800/1*r1N9X-eFEUDm2FSibwdJ3g.jpeg) 60 | 61 | 来源: 62 | 63 | 我绝对肯定许多人工智能研究人员会对这项监管提案表示支持。但是，您需要能够证明面部识别功能的好处绝对大于风险。风险是非常真实的，现在是我们开始认真讨论的时候了。 64 | 65 | **进一步阅读** 66 | 67 | [**面部识别软件尚未准备好供执法部门使用**](https://techcrunch.com/2018/06/25/facial-recognition-software-is-not-ready-for-use-by-law-enforcement/) 68 | 69 | 70 | 71 | ![img](https://cdn-images-1.medium.com/max/800/1*j9kar_3vwdJK8twhtmXC0g.png) 72 | 73 | 更多报道： -------------------------------------------------------------------------------- /2017年10月/20171015 第13期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---------------------------- | 3 | | [PyTorch tutorial distilled](https://towardsdatascience.com/pytorch-tutorial-distilled-95ce8781a89c?from=hackcv&hmsr=hackcv.com) | PyTorch简单的介绍 | 4 | | [Intro To Data Analysis For Everyone! Part 1](https://towardsdatascience.com/intro-to-data-analysis-for-everyone-part-1-ff252c3a38b5?from=hackcv&hmsr=hackcv.com) | 数据科学家需要具备的几个能力 | 5 | | [The Search for Better Search at Reddit](https://redditblog.com/2017/09/07/the-search-for-better-search-at-reddit/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | Reddit的搜索功能实现介绍 | 6 | -------------------------------------------------------------------------------- /2017年10月/20171016 第14期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---- | 3 | | [BUILDING A NEURAL NET FROM SCRATCH IN GO](https://www.datadan.io/building-a-neural-net-from-scratch-in-go/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 介绍用Go构建一个简单的神经网络 | 4 | | [A Quick Introduction to Neural Networks](https://ujjwalkarn.me/2016/08/09/quick-intro-neural-networks/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 神经网络的快速入门 | 5 | | [Practical Data Science in Python](https://radimrehurek.com/data_science_python/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 用垃圾邮件分类完整的介绍了建立机器学习模型过程（Python2) | -------------------------------------------------------------------------------- /2017年10月/20171017 第15期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---- | 3 | | [DISTRIBUTED ALGORITHMS IN NOSQL DATABASES](https://highlyscalable.wordpress.com/2012/09/18/distributed-algorithms-in-nosql-databases/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 详细的介绍了分布式的一些实现原理 | 4 | | [MAPREDUCE PATTERNS, ALGORITHMS, AND USE CASES](https://highlyscalable.wordpress.com/2012/02/01/mapreduce-patterns/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 介绍了MapReduce的原理及应用并附上伪代码 | 5 | | [Machine learning skills for software engineers](https://www.infoworld.com/article/3223688/machine-learning/machine-learning-skills-for-software-engineers.html?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 对于软件工程师来说，是需要学习一些基础知识的 | -------------------------------------------------------------------------------- /2017年10月/20171018 第16期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---- | 3 | | [Benefits of Intel® Optimized Caffe* in comparison with BVLC Caffe*](https://software.intel.com/en-us/articles/comparison-between-intel-optimized-caffe-and-vanilla-caffe-by-intel-vtune-amplifier?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 4 | | [NEW OPTIMIZATIONS IMPROVE DEEP LEARNING FRAMEWORKS FOR CPUS](https://www.nextplatform.com/2017/10/13/new-optimizations-improve-deep-learning-frameworks-cpus/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 5 | | [Why SQL is beating NoSQL, and what this means for the future of data](Why SQL is beating NoSQL, and what this means for the future of data) | | -------------------------------------------------------------------------------- /2017年10月/20171019 第17期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---- | 3 | | [A Tour of Gotchas When Implementing Deep Q Networks with Keras and OpenAi Gym](http://srome.github.io/A-Tour-Of-Gotchas-When-Implementing-Deep-Q-Networks-With-Keras-And-OpenAi-Gym/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 4 | | [Scientists Can Read a Bird’s Brain and Predict Its Next Song](https://www.technologyreview.com/s/609032/scientists-can-read-a-birds-brain-and-predict-its-next-song/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 5 | | [TensorFlow* Optimizations on Modern Intel® Architecture](https://software.intel.com/en-us/articles/tensorflow-optimizations-on-modern-intel-architecture?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | -------------------------------------------------------------------------------- /2017年10月/20171019 第17期/科学家可以读取一只鸟的大脑并且预测它的下一声啼叫.md: -------------------------------------------------------------------------------- 1 | # 科学家可以读取一只鸟的大脑并且预测它的下一声啼叫 2 | 3 | 原文链接：[Scientists Can Read a Bird’s Brain and Predict Its Next Song](https://www.technologyreview.com/s/609032/scientists-can-read-a-birds-brain-and-predict-its-next-song/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) 4 | 5 | ## 下一步，通过一个脑机接口预测人类的讲话。 6 | 7 | ​ 在硅谷的企业家们今年给自己设定了一个大胆的新目标：创造一个读脑设备使人们可以通过他们的思想毫不费力的传递信息。 8 | 9 | ​ 在四月，埃隆 · 马斯克爆出了一家叫做“[Neuralink](https://www.technologyreview.com/s/604254/with-neuralink-elon-musk-promises-human-to-human-telepathy-dont-believe-it/) ”的人脑接口新公司。几天后，脸书的CEO马克 · 扎克伯格宣称：“可用的人脑接口即将到来，最终，让你可以只用你的思想进行交流。”脸书表示有[六十名工程师](https://www.technologyreview.com/s/604229/facebooks-sci-fi-plan-for-typing-with-your-mind-and-hearing-with-your-skin/)正在研究这个问题。 10 | 11 | ​ 这是一个有野心的探索，并且我们有理由认为它不会很快的就发生。但至少对一只橙色喙的小鸟——斑胸草雀来说，这个梦想离现实更近了。 12 | 13 | ​ 这要感谢Timothy Gentner 和他在加利福尼亚大学学生们的杰出工作，他们设计了一个“大脑到鸣叫”的界面，在此界面中，他可以预测一只雀在几秒钟之后的啼叫。 14 | 15 | ​ “我们直接从神经行为解码现实中各种鸟叫”这是科学家发表在网站 [bioRxiv](https://www.biorxiv.org/content/early/2017/09/27/193987) 的一篇新报告中提到的。阿根廷鸟鸣专家Ezequiel Arneodo 所在的团队称这个系统是一个从神经行为到复杂的神经交流信号的解码器的原型。研究人员说，类似的方法可以推动人类思想到文字转换接口的发展。 16 | 17 | ​ 科学家们说他们可以直接从一只雀的大脑活动预测这只雀的叫声。 18 | 19 | ​ 一只鸣禽的大脑并不大。但它的发声方式与人类的语言相似，这使得这些鸟成为研究记忆和认知的科学家们的最爱。它们的叫声很复杂，就像人类的语言一样，它们也是后天习得的。斑胸草雀从一只年长的鸟那里学会了它的叫声。 20 | 21 | ​ Makoto Fukushima是美国国立卫生研究院的一名研究员，他使用大脑接口来研究猴子发出的较简单的咕哝声和咕咕声。他说，鸟鸣声的范围更广，这就是为什么新的研究结果“对人类语言的应用具有重要意义”。 22 | 23 | ​ 目前在人类中尝试的大脑界面主要是跟踪反映一个人想要手臂做出特定活动的神经信号，这种神经信号可以用来移动机器人或引导光标非常缓慢地啄出字母。因此，那种可以毫不费力接收到你想要说的内容的头盔或大脑植入装置还远未实现。 24 | 25 | ​ 但正如这项新研究显示的那样，这并非完全不可能。加州大学圣地亚哥分校的研究小组利用醒着的鸟儿身上的硅电极，测量了大脑中被称为感觉运动核的部分神经元的电颤，而这部分神经元正是“形成学习歌曲的指令”产生的地方。 26 | 27 | ​ 这项实验使用了神经网络软件，这是一种机器学习方法。研究人员在程序中输入神经放电的模式和实际产生的叫声，以及它的停止和开始和频率的变化。他们的想法是训练他们的软件去匹配其中一个他们称之为“神经到歌曲的频谱映射”。 28 | 29 | ​ 该团队的主要创新之处在于，通过整合雀科鸣叫的物理模型，简化了从大脑到啼叫的翻译。鸟类不像人类那样有声带。相反，它们把空气喷到喉咙中振动的表面上，这种表面被称为鸣管 。想想看，如果你把两张纸放在一起，在纸的边缘吹气，就能发出尖锐的呜呜声。 30 | 31 | ​ 总结：作者说:“我们直接从神经活动中解码现实中各种各样鸟鸣。在他们的报告中，研究小组称，他们可以在一只鸟发声的30毫秒之前预测它会如何啼叫。 32 | 33 | ​ 你可以在下面的音频中听到结果。记住斑马雀不是夜莺。它的歌更像是断断续续的嘎嘎叫。 34 | 35 | ​ 结果证实是同一种啼叫，正如从雀脑内的神经记录预测的那样。 36 | 37 | ​ 鸣禽已经是一个重要的研究模型。在埃隆·马斯克的神经网络学院(Neuralink)，鸟类科学家是首批聘用的关键人员之一。UCSD专注检测的言语背后肌肉运动技巧也可能是脑机接口技术的一个关键突破点。 38 | 39 | ​ Facebook表示，它希望人们能够以每分钟100字的速度直接通过大脑打字，随时随地都可以私下发短信。当你需要用唇语表达你的意思时，一个能读懂大脑对肌肉发出的指令的设备可能比一个能读懂“思想”的设备要现实得多。 40 | 41 | ​ Gentner和他的团队希望他们的雀类能够帮助实现这一目标。他们写道:“我们已经用动物模型演示了一种复杂的通信信号(脑机接口)，”。还补充说，“我们的方法也为生物医学语音假体设备提供了一个有价值的试验场。” 42 | 43 | ​ 换句话说，我们离通过大脑发短信的目标又近了一小步。 -------------------------------------------------------------------------------- /2017年10月/20171020 第18期/AlphaGoZero从零开始学习.md: -------------------------------------------------------------------------------- 1 | # AlphaGo Zero: 从零开始学习 2 | 3 | 原文链接：[AlphaGo Zero: Learning from scratch](https://deepmind.com/blog/alphago-zero-learning-scratch/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) 4 | 5 | 人工智能研究在语音识别和图片分类到基因学和药物发现等不同的领域取得了巨大的进步。在许多场景中，都有大量利用人们专业之和数据的专业的系统 6 | 7 | 然而，对于一些问题来说，利用人们的知识也许代价太大，太过不可靠，或者根本不可用。因此，人工智能长期的目标就是跨过这一步——创造在没有人为输入的情况下可以在绝大多数充满挑战性的领域里展现出超人一般的表现的算法。在[《自然杂志》](https://www.nature.com/)上发表的最新[论文](http://nature.com/articles/doi:10.1038/nature24270)中，我们展示了向这个目标迈出的重要一步。 8 | 9 | # 从零开始 10 | 11 | ![img](https://storage.googleapis.com/deepmind-live-cms/images/AlphaGoZero-Illustration-WideScreen.width-320_oOByzmR.jpg) 12 | 13 | 这篇论文介绍了最新一代的AlphaGo产品AlphaGo Zero，第一个在古中国下棋游戏中打败了世界冠军的电脑程序。AlphaGo Zero甚至可以说是能力最强的并且按理说是历史上最强的下棋选手。 14 | 15 | 以前版本的AlpGo最初与成千上万的人类业余爱好者和专业游戏者训练来学习如何下棋。AlphaGoZero跳过了这一步骤并通过和自己下棋来学习，从最开始的胡乱下棋开始。通过这个方法，它很快的就超越了人类下棋的水平，并且以100比0的战绩打败了[之前发布](http://www.nature.com/nature/journal/v529/n7587/full/nature16961.html?foxtrotcallback=true)的打败了下棋冠军的AlpGo版本。 16 | 17 | ![Training time graphic](https://storage.googleapis.com/deepmind-live-cms/documents/TrainingTime-Graph-171019-r01.gif) 18 | 19 | 之所以能够这么做是因为它使用了[强化学习](https://en.wikipedia.org/wiki/Reinforcement_learning)的一种新的形式——AlpGoZero变成了他自己的老师。这个系统由一个完全不知道下棋游戏的神经网络系统开始。然后，通过将这个神经网络系统和强大的搜索算法相结合，他就会他自己和自己下棋。随着它不断地和自己下棋，神经网络不断被调整和升级来预测下一步动作，并最终成为游戏的赢家。 20 | 21 | 这个被更新过的神经网络之后又和搜索算法结合来创造新的更加强大版本的AlpGo Zero，并且这个过程还会再次开始。在每次迭代中这个系统的表现都会有微小的提升，并且自己玩游戏的质量也会有所增加，导致越来越多的准确的神经网络和甚至更加强大版本的AlphaGoZero。 22 | 23 | 这个技术比之前版本的AlpGo更加强大因为它不再局限于人们现有的知识。取而代之的是他能够从零开始向世界上最强的下棋选手:AlpGo他自己学习。 24 | 25 | 它还有其他一些有别于之前版本的显著的区别： 26 | 27 | - AlpGoZero仅仅将棋盘上的黑棋白棋作为输入，而之前版本的AlpGo则是包括了少量的手工设计的特性 28 | - 他使用一个神经网络而不是两个，之前版本的AlpGo使用的是policy network来选择下一步做什么然后使用value network来预测游戏赢家的每一个位置。这两个神经网络在AlpGoZero中被结合了，使得它能够更加有效的去训练和评估。 29 | - AlphaGo Zero不使用“初赛”——其他围棋程序使用快速、随机的游戏来预测哪位棋手将在目前的棋盘位置上获胜。取而代之的它依赖于他自己高质量的神经网络来评估位置 30 | 31 | 所有这些差异都有助于提高系统的性能，使其更加通用。但是这正是算法的改变才使得系统更加的强力和高效。 32 | 33 | ![img](https://storage.googleapis.com/deepmind-live-cms/images/AlphaGo%2520Efficiency.width-400_cHoMue6.png) 34 | 35 | AlphaGo的效率越来越高得益于硬件的进步和算法的优化。 36 | 37 | 经过三天的自我游戏训练，AlphaGo Zero以100比0的比分击败了之前发布的[AlphaGo](https://research.googleblog.com/2016/01/alphago master -ancient-game-of-go.html)。经过40天的自我训练，AlphaGo Zero变得更加强大，超过了被称为“Master”的AlphaGo版本，后者击败了世界上最好的选手和世界第一的棋手[柯洁](https://deepmind.com/research/alphago/alphago-china/)。 38 | 39 | ![img](https://storage.googleapis.com/deepmind-live-cms/images/Elo%2520Ratings.width-400_ahXVKga.png) 40 | 41 | Elo评定-围棋等竞技游戏中玩家相对技能水平的的衡量方法-展示处理AlphaGo如何在它的发展中变得强大的。 42 | 43 | AlphaGo在数以百万计的和自己比赛的过程中，系统逐渐地从零开始学习围棋，在短短几天内积累了数千年的人类知识。AlphaGo也发现了新的知识，自创了非传统的策略和创造性的新行为。与李世石(Lee Sedol)和柯洁(Ke Jie)的比赛中使用的新技术相呼应，并超越了后者。 44 | 45 | ![AlphaGo Zero knowledge timeline](https://storage.googleapis.com/deepmind-live-cms/documents/Knowledge%2520Timeline.gif) 46 | 47 | 这些创造性的时刻给了我们信心——人工智能将是人类创造力的倍增器，帮助我们完成我们自己的[任务](https://deepmind.com/about/)以解决人类正在面临的一些重要挑战 48 | 49 | 50 | 51 | # 发现新的知识 52 | 53 | ![img](https://storage.googleapis.com/deepmind-live-cms/images/AlphaGoZero-Illustration-Square.width-320_RDH0108.jpg) 54 | 55 | 虽然现在还为时尚早，但AlphaGo Zero是迈向这一目标的关键一步。如果类似的技术可以应用于其他结构性问题，如蛋白质折叠、降低能源消耗或寻找革命性的新材料，那么由此产生的突破有可能对社会产生积极影响。 56 | 57 | ------ 58 | 59 | 阅读 [这篇文章](https://www.nature.com/articles/nature24270.epdf?author_access_token=VJXbVjaSHxFoctQQ4p2k4tRgN0jAjWel9jnR3ZoTv0PVW4gB86EEpGqTRDtpIz-2rmo8-KG06gqVobU5NSCFeHILHcVFUeMsbvwS-lxjqQGg98faovwjxeTUgZAUMnRQ) 60 | 61 | 阅读这篇文章相关的 [Nature News and Views article](https://www.nature.com/articles/550336a.epdf?shared_access_token=QbXlOw9nSIP_MS1moc_M0tRgN0jAjWel9jnR3ZoTv0PvinEKRXS2Dk736vL8i-Uo2-6AN8KRxOlLhDGorUgFzEgC3fwrX95r3LQ7u2FBwQ5axjmpMSZrWg4i6D7_g5rV5ze0zLhgo4jufsSKL-UZmw%3D%3D) 62 | 63 | 下载 [AlphaGo Zero games](http://www.alphago-games.com/) 64 | 65 | 阅读 [更多AlpGo的文章](https://deepmind.com/research/alphago/) 66 | 67 | **\*这个作品是由David Silver, Julian Schrittwieser, Karen Simonyan, Ioannis Antonoglou, Aja Huang, Arthur Guez, Thomas Hubert, Lucas Baker, Matthew Lai, Adrian Bolton, Yutian Chen, Timothy Lillicrap, Fan Hui, Laurent Sifre, George van den Driessche, Thore Graepel和Demis Hassabis完成的*** -------------------------------------------------------------------------------- /2017年10月/20171020 第18期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---- | 3 | | [Perceptually uniform color spaces](https://programmingdesignsystems.com/color/perceptually-uniform-color-spaces/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 4 | | [Stop Using word2vec](https://multithreaded.stitchfix.com/blog/2017/10/18/stop-using-word2vec/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 5 | | [AlphaGo Zero: 从零开始学习](https://deepmind.com/blog/alphago-zero-learning-scratch/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 6 | 7 | -------------------------------------------------------------------------------- /2017年10月/20171021 第19期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---- | 3 | | [How to Prepare Movie Review Data for Sentiment Analysis](https://machinelearningmastery.com/prepare-movie-review-data-sentiment-analysis/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 4 | | [8 Essential Tips for People starting a Career in Data Science](https://www.analyticsvidhya.com/blog/2017/10/tips-people-starting-career-data-science/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 5 | | [Hey Siri: An On-device DNN-powered Voice Trigger for Apple’s Personal Assistant](https://machinelearning.apple.com/2017/10/01/hey-siri.html?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | -------------------------------------------------------------------------------- /2017年10月/20171022 第20期/AlphaGo与能力增强.md: -------------------------------------------------------------------------------- 1 | # AlphaGo Zero与能力增强 2 | 3 | 原文链接：[AlphaGo Zero and capability amplification](https://ai-alignment.com/alphago-zero-and-capability-amplification-ede767bb8446?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) 4 | 5 | [AlphaGo Zero](https://deepmind.com/blog/alphago-zero-learning-scratch/) 是AI能力的一个很好的演示，也是一个很好的观点证明了一个有效的对应策略。 6 | 7 | ### AlphaGo Zero是如何工作的 8 | 9 | AlphaGo Zero 通过两个函数进行学习（作为当前方式的输入） 10 | 11 | - 一个超前的过度动作p是被训练来预测AlphaGo 最终落子位置的。 12 | - 一个价值函数v是被训练来预测哪一位选手会赢。（如果是AlpahaGo 自己和自己下。） 13 | 14 | 这两个训练都是监督学习。一旦我们有了这两个函数，AlphaGo 将用p和v去判断Monte Carlo树的1600步得到的结果。通过这样高代价的搜索过程来训练p，使p最后有一个好的移动表现。随着p的增强，高代价的搜索就变得越来越强大，而p就越来越趋向于理想化的那个值。 15 | 16 | ### 迭代能力增强 17 | 18 | 在[迭代能力增强最简单的形式](https://ai-alignment.com/benign-model-free-rl-4aae8c97e385)中,我们训练了一个函数： 19 | 20 | - 一个“弱”策略A，它被训练来预测在给定的情况下代理最终会决定什么。 21 | 22 | 就像AlphaGo 不会用p直接决定移动，我们也不会用弱策略直接决定行动。我们会用能力增强的方案替代：我们多次调用A以取得更加智能的判断。我们训练A去忽视高代价的增强过程，为了直接取得智能的判断。随着A的增强，放大的策略变得更加强大，A不断趋向于理想化的那个值。 23 | 24 | 以AlphaGo Zero举例，A是超前的过度动作，增强策略是MCTS。（确切的说，A是一对（p,v)，而能力增强厕所是MCTS加上使用一个滚动查看谁会赢。） 25 | 26 | 抛开AlphaGo Zero 不说，A可以看作是一个问题回答系统，它在很多时候被用于将一个问题分解成各个部分后分散解决。又或者可以把它看成是一个认知工作空间的[更新策略](https://blog.ought.com/dalca-4d47a90edd92)，很多时候可以用于使解决问题时想的更远。 27 | 28 | ### 意义 29 | 30 | 加强学习者用激励函数去优化它，但不幸的是，我们并不知道在哪里用激励函数可以使它忠实的追踪我们所关心的事。这是安全问题的一个重要来源。 31 | 32 | 相比之下，AlphaGo Zero 采取策略改进操作（比如MCTS）并收敛到它的不动点。如果我们能够找到一个方法去改进这个策略的同时使其不变，那么我们可以应用同样的算法来获得非常强大但整齐的策略。 33 | 34 | 使用MCTS在现实世界中去实现一个简单的目标都无法使其对齐，因此它不符合要求。但是在“[想的更远](https://ai-alignment.com/humans-consulting-hch-f893f6051455)”的方面是可能做到的。只要我们以一个[足够接近](https://ai-alignment.com/corrigibility-3039e668638)对其的政策开始—“想要“使其对齐，从某种意义上说，允许它思考的远一些就可能使它更聪明也更一致。 35 | 36 | 我认为在如今设计保持对其的方法增强使一个容易处理的问题，可以在现有ML或人工调整的上下文中对其进行研究。所以我认为这使一个在AI校准方面一个可以一试的方向。后备方案可以直接合并到AlphaGo Zero的体系结构中，所以我们已经可以获得相关经验的反馈。如果运气好的话，表现好的AI系统会表现得像AlphaGo Zero,这就可以给我们很多方法去校准AI。 37 | 38 | -------------------------------------------------------------------------------- /2017年10月/20171022 第20期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ------------------------------ | 3 | | [A Neural Network in 11 lines of Python (Part 1)](http://iamtrask.github.io/2015/07/12/basic-python-network/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 一个简单的模型并详细解释了。 | 4 | | [AlphaGo Zero and capability amplification](https://ai-alignment.com/alphago-zero-and-capability-amplification-ede767bb8446?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 关于Alphago_zero的一些分析。 | 5 | | [Improving Real-Time Object Detection with YOLO](https://blog.statsbot.co/real-time-object-detection-yolo-cd348527b9b7?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | 目标检测问题的YOLO的相关介绍。 | 6 | 7 | -------------------------------------------------------------------------------- /2017年10月/20171022 第20期/使用YOLO提高实时目标检测能力.md: -------------------------------------------------------------------------------- 1 | # 使用YOLO提高实时目标检测能力 2 | 3 | 原文链接：[Improving Real-Time Object Detection with YOLO](https://blog.statsbot.co/real-time-object-detection-yolo-cd348527b9b7?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) 4 | 5 | ## 一个关于实时目标检测的新观点 6 | 7 | 近年来，在深度学习的帮助下，目标检测领域取得了巨大的进展。目标检测是在一个图片中识别物体并在物体周围画一个框标出来，即确定它们的位置。这在计算机视觉邻域是一个十分重要的问题，因为它经常应用于自动驾驶汽车的安全和追踪的应用上。 8 | 9 | 早先的目标检测方法通常是将图片按一定顺序分离成不同的部分，这就导致完成的每一部分和最后的结果没有连接，就是将图像中的对象画一个框包围。一个端对端框架在降低一个联合的图中的识别错误方面是一个很好的加爵办法，它不仅是训练模型使其有更高的准确率，还可以提高它的识别速度。 10 | 11 | 这就是You Only Look Once（YOLO）算法可以应用的场景。Varun Agrawa 告诉[Statsbot](https://statsbot.co/?utm_source=blog&utm_medium=article&utm_campaign=yolo)的团队为什么YOLO比别的算法更适合在目标检测问题上应用。 12 | 13 | ![img](https://cdn-images-1.medium.com/max/2000/1*PSFl5og1c9HIKXlMIJV8-Q.png) 14 | 15 | [Illustration source](https://arxiv.org/abs/1506.02640) 16 | 17 | 深度学习已经被证明了使图像分类问题最有效的工具，并已经在这个问题上具备了人工级别的能力。早期的检测方法利用这种能力将目标检测转换为分类问题之一，即识别图像目标属于哪种类型的目标。 18 | 19 | 这样的方式是通过2个步骤完成的： 20 | 21 | 1.第一步包括产生数以万计的想法，他们都只是图像上特定的矩形区域，也被称为边界框。边界框可以围绕图像中的对象进行，也可以不。而对这些想法进行过滤是第二步的目标。 22 | 23 | 2.第二步是一个图像分类器会对边界框内包括的子图像进行分类，判断它是否是一个特别的目标类型的一部分又或者只是简单的无目标的或者只是背景。 24 | 25 | 这两步过程虽然非常精确，但是仍然存在一些缺陷，例如效率问题，这是由于生成了大量的想法，并且缺乏对想法生成和分类的联合优化。这导致每个阶段都不能真正理解全局，而只是被困在自己的小问题中，从而限制了他们整体的性能。 26 | 27 | ### **YOLO到底是什么** 28 | 29 | 这就是YOLO引入的地方。YOLO代表着你只看一次，是一种基于目标识别算法的深度学习。在2016年由华盛顿大学的 [Joseph Redmon and Ali Farhadi](https://arxiv.org/abs/1506.02640) 开发出来的。 30 | 31 | YOLO系统背后的理论基础不再是过去的那种传递多个可能的子图片，而是仅向深度学习系统传递一次完整的图片。接着，你将会得到所有的边界框以及目标类别分类。这是YOLO的基本设计策略，这是在目标识别领域的一个崭新的视角。 32 | 33 | YOLO工作的方式是将图像细分成NXN网格，或者更具体地说，在原始文件7x7网格中的每个网格单元，也称为锚，表示一个分类器，它负责在潜在对象的周围生成K个边界框，潜在对象的基本真值中心落在该网格单元内（本文中K是2），并将其分类为正确的对象。 34 | 35 | > *注意，边界框不限于网格单元内，它可以在图像的边界内扩展，以容纳它认为负责检测的对象。这意味着在当前版本的YOLO中，系统生成98个不同大小的边界框，以适应场景中的各种对象。* 36 | 37 | ### 性能和结果 38 | 39 | 对于更密集的对象检测，用户可以根据他们的需要将K或N设 置为更高的数字。然而，在当前的配置中，我们有一个系统，该系统能够输出大量围绕对象的边界框，并且基于图像的空间布局将它们分类到各种类别。 40 | 41 | 在运行时间里只有一次通过图像。因此，联合检测和分类可以更好地优化学习目标（损失函数）和实时性能。 42 | 43 | 事实上，YOLO的结果是非常理想的。在具有挑战性的[Pascal VOC检测挑战数据集](http://host.robots.ox.ac.uk/pascal/VOC/pubs/everingham15.pdf)上，YOLO以每秒45帧的速度运行时，实现了63.4（每100帧中）的平均精度或mAP。相比之下，根据现有模型，Faster-RCNN VGG 16实现了73.2的mAP ，但是仅以最大每秒7帧的速度运行，效率降低了6倍。 44 | 45 | 你可以在下面的图表中看到YOLO与其他模型的不同。 46 | 47 | ![img](https://cdn-images-1.medium.com/max/1600/1*rZR8fU2sIz2DSIJqkBb4iA.png) 48 | 49 | > *YOLO在牺牲一些精度的条件下，它可以以每秒155帧运行，虽然只有在52.7的mAP。* 50 | 51 | 因此，YOLO的主要优势是其在实时速度下的目标检测中稳定的良好性能。这使它能在机器人、自动驾驶汽车和无人机等系统中使用，在这些系统中，时间至关重要。 52 | 53 | ### YOLOv2 的框架 54 | 55 | 最近，同样的研究人员发布了新的YOLOv2框架，该框架利用最近在深层学习网络设计中得到的结果来构建更有效的网络，并且使用Faster-RCNN的锚箱思想来减轻网络的学习问题。 56 | 57 | ![img](https://cdn-images-1.medium.com/max/1200/0*X3S2jCdO6bcgCdyc.) 58 | 59 | [插图来源](http://www.pjreddie.com/) 60 | 61 | YOLOv2有着更好的检测结果，在Pascal VOC检测数据集上以78.6mAP实现最先进的性能，而其他系统，例如改进的Faster-RCNN(Faster-RCNN ResNet)和[SSD500](https://arxiv.org/pdf/1512.02325.pdf)，在相同的测试数据上仅实现76.4mAP和76.8mAP。 62 | 63 | > *关键的区别是性能速度。性能最好的YOLVO2模型运行在40 FPS相比5 FPS的 FAST-RCNN RESNET* 64 | 65 | 虽然 SSD500以45FPS运行，但是具有mAP 76.8（与SSD500相同）的低分辨率版本的YOLOv2在67FPS下运行，由于YOLOv2的设计选择，这向我们展示了YOLOv2的高性能能力。 66 | 67 | ### 最后的一些感想 68 | 69 | 综上所述，YOLO在实时性能上运行时表现出了显著的性能增益，这是在资源匮乏的深度学习算法时代一个重要的中间环节。随着我们向着更加自动化的未来迈进，像YOLO和SSD500这样的系统已经准备好迎来大步的进步，逐步实现伟大的AI梦想。 70 | 71 | ### 一些重要的阅读文章 72 | 73 | - [You Only Look Once: Unified, Real-Time Object Detection](https://arxiv.org/abs/1506.02640) 74 | - [The PASCAL Visual Objects Challenge: A Retrospective](http://host.robots.ox.ac.uk/pascal/VOC/pubs/everingham15.pdf) 75 | - [SSD: Single Shot Multibox Detector](https://arxiv.org/pdf/1512.02325.pdf) 76 | 77 | -------------------------------------------------------------------------------- /2017年10月/20171023 第21期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---- | 3 | | [Intel® Nervana™ Neural Network Processors (NNP) Redefine AI Silicon)](http://iamtrask.github.io/2015/07/12/basic-python-network/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 4 | | [Bayesian Learning for Statistical Classification](https://blog.statsbot.co/bayesian-learning-for-statistical-classification-f2362d620428?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 5 | | [Instacart Market Basket Analysis, Winner's Interview: 2nd place, Kazuki Onodera](http://blog.kaggle.com/2017/09/21/instacart-market-basket-analysis-winners-interview-2nd-place-kazuki-onodera/?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 6 | 7 | -------------------------------------------------------------------------------- /2017年10月/20171024 第22期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | ------------------------------------------------------------ | ---- | 3 | | [Word embeddings in 2017: Trends and future directions](http://ruder.io/word-embeddings-2017/index.html?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 4 | | [Learning Maths for Machine Learning and Deep Learning](https://medium.com/towards-data-science/learning-maths-for-machine-learning-and-deep-learning-5509c097ee83?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 5 | | [Advice For New and Junior Data Scientists](https://medium.com/@rchang/advice-for-new-and-junior-data-scientists-2ab02396cf5b?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 6 | 7 | -------------------------------------------------------------------------------- /2017年10月/20171024 第22期/Word embeddings in 2017 Trends and future directions.md: -------------------------------------------------------------------------------- 1 | # Learning Maths for Machine Learning and Deep Learning 2 | 3 | 原文链接：[Learning Maths for Machine Learning and Deep Learning](https://medium.com/towards-data-science/learning-maths-for-machine-learning-and-deep-learning-5509c097ee83?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) 4 | 5 | While I did learn a lot of maths while doing my engineering degree, I forgot most of it by the time I wanted to get into Machine Learning. After I graduated I never really had a need for any of the maths. I did a lot of web programming which relied on logic and I can honestly say that with each system with the word ‘Management’ in the title I lost a third of my math knowledge! I’ve programmed extensions for Learning Management Systems, Content Management Systems and Customer Relationship Management Systems — I’ll leave you to figure out how much math apptitude I had after working with these systems. At the moment I’ve got good data science skills and can use a variety of ML and DL algorithms. I’ve successfully completed a number of MOOCs (e.g., Deep Learning Foundations from Udacity and Andrew Ng’s new Coursera courses). I can use Scikit Learn, TensorFlow and Kera’s. …. but I have rough ideas for creating new variants of algorithms. At the moment I really want to create a new kind of interactive topic modeling algorithm. I’ve felt stuck due to my lack of maths knowledge. In my travels trying to re-learn some basic maths, I’ve come across a couple of books that have been written by people with the art of explanation. These books have made a tremendous difference as they are able to convey complex concepts in a very simple manner. I am writing this blog post to share these great resources especially for programmers. The books cover Calculus and Linear Algebra. I’ve not found an equivalent Probability and Statistics book yet — If you know of one please leave a comment or tweet [Aneesha Bakharia](https://medium.com/@aneesha). 6 | 7 | ### Calculus Made Easy by S. Thompson 8 | 9 | Learn calculus from a book written in 1914! The [pdf for the book](http://djm.cc/library/Calculus_Made_Easy_Thompson.pdf) is freely available. This book is simply amazing. The English is a bit old style but the explanations are timeless. Thompson makes calculus super easy. Optimization of a cost function is core to ML and DL and this book will help you understand the basics of minimization. Those update rules in gradients decent won’t seem like magic anymore. Just read the prologue — its set the tone for the rest of the book… 10 | 11 | 12 | 13 | ![img](https://cdn-images-1.medium.com/max/1000/1*idSk5MAMQC7eNzg29PqOIQ.png) 14 | 15 | The Prologue from Calculus Made Easy by S. Thompson 16 | 17 | ### Coding the Matrix by P. N. Klein 18 | 19 | Most Linear Algebra books start easy but then concepts like image, basis, dimension, orthogonalization, eigenvectors are introduced in a completely abstract way. Most Linear Algebra books fail to even introduce real world applications and its hard to see where or why you would use the math. Matrix multiplication is a good example of something I learnt but never truely understood (i.e., why is was not performed element by element). [Coding the Matrix](http://codingthematrix.com/) is different! You actually get to build your own linear algebra library while improving your Python programming skills! The book is full of practical computer science applications (e.g. fix the perspective of a whiteboard photograph). 20 | 21 | 22 | 23 | ![img](https://cdn-images-1.medium.com/max/1000/1*RshBwDdVw5oGoFVRgh-y-w.png) 24 | 25 | Coding the Matrix — the best Linear Algebra book ever! -------------------------------------------------------------------------------- /2017年10月/20171025 第23期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年10月/20171025 第23期/README.md -------------------------------------------------------------------------------- /2017年10月/20171026 第24期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年10月/20171026 第24期/README.md -------------------------------------------------------------------------------- /2017年10月/20171027 第25期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年10月/20171027 第25期/README.md -------------------------------------------------------------------------------- /2017年10月/20171028 第26期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年10月/20171028 第26期/README.md -------------------------------------------------------------------------------- /2017年10月/20171029 第27期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年10月/20171029 第27期/README.md -------------------------------------------------------------------------------- /2017年10月/20171030 第28期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年10月/20171030 第28期/README.md -------------------------------------------------------------------------------- /2017年10月/20171031 第29期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年10月/20171031 第29期/README.md -------------------------------------------------------------------------------- /2017年11月/20171101 第30期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171101 第30期/README.md -------------------------------------------------------------------------------- /2017年11月/20171102 第31期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171102 第31期/README.md -------------------------------------------------------------------------------- /2017年11月/20171103 第32期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171103 第32期/README.md -------------------------------------------------------------------------------- /2017年11月/20171104 第33期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171104 第33期/README.md -------------------------------------------------------------------------------- /2017年11月/20171105 第34期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171105 第34期/README.md -------------------------------------------------------------------------------- /2017年11月/20171106 第35期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171106 第35期/README.md -------------------------------------------------------------------------------- /2017年11月/20171107 第36期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171107 第36期/README.md -------------------------------------------------------------------------------- /2017年11月/20171108 第37期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171108 第37期/README.md -------------------------------------------------------------------------------- /2017年11月/20171109 第38期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171109 第38期/README.md -------------------------------------------------------------------------------- /2017年11月/20171110 第39期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171110 第39期/README.md -------------------------------------------------------------------------------- /2017年11月/20171111 第40期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171111 第40期/README.md -------------------------------------------------------------------------------- /2017年11月/20171112 第41期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171112 第41期/README.md -------------------------------------------------------------------------------- /2017年11月/20171113 第42期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171113 第42期/README.md -------------------------------------------------------------------------------- /2017年11月/20171114 第43期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171114 第43期/README.md -------------------------------------------------------------------------------- /2017年11月/20171115 第44期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171115 第44期/README.md -------------------------------------------------------------------------------- /2017年11月/20171116 第45期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171116 第45期/README.md -------------------------------------------------------------------------------- /2017年11月/20171117 第46期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171117 第46期/README.md -------------------------------------------------------------------------------- /2017年11月/20171118 第47期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171118 第47期/README.md -------------------------------------------------------------------------------- /2017年11月/20171119 第48期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171119 第48期/README.md -------------------------------------------------------------------------------- /2017年11月/20171120 第49期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171120 第49期/README.md -------------------------------------------------------------------------------- /2017年11月/20171121 第50期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171121 第50期/README.md -------------------------------------------------------------------------------- /2017年11月/20171122 第51期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171122 第51期/README.md -------------------------------------------------------------------------------- /2017年11月/20171123 第52期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171123 第52期/README.md -------------------------------------------------------------------------------- /2017年11月/20171124 第53期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171124 第53期/README.md -------------------------------------------------------------------------------- /2017年11月/20171125 第54期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171125 第54期/README.md -------------------------------------------------------------------------------- /2017年11月/20171126 第55期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171126 第55期/README.md -------------------------------------------------------------------------------- /2017年11月/20171127 第56期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171127 第56期/README.md -------------------------------------------------------------------------------- /2017年11月/20171128 第57期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171128 第57期/README.md -------------------------------------------------------------------------------- /2017年11月/20171129 第58期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171129 第58期/README.md -------------------------------------------------------------------------------- /2017年11月/20171130 第59期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年11月/20171130 第59期/README.md -------------------------------------------------------------------------------- /2017年12月/20171201 第60期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171201 第60期/README.md -------------------------------------------------------------------------------- /2017年12月/20171202 第61期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171202 第61期/README.md -------------------------------------------------------------------------------- /2017年12月/20171203 第62期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171203 第62期/README.md -------------------------------------------------------------------------------- /2017年12月/20171204 第63期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171204 第63期/README.md -------------------------------------------------------------------------------- /2017年12月/20171205 第64期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171205 第64期/README.md -------------------------------------------------------------------------------- /2017年12月/20171206 第65期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171206 第65期/README.md -------------------------------------------------------------------------------- /2017年12月/20171207 第66期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171207 第66期/README.md -------------------------------------------------------------------------------- /2017年12月/20171208 第67期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171208 第67期/README.md -------------------------------------------------------------------------------- /2017年12月/20171209 第68期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171209 第68期/README.md -------------------------------------------------------------------------------- /2017年12月/20171210 第69期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171210 第69期/README.md -------------------------------------------------------------------------------- /2017年12月/20171211 第70期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171211 第70期/README.md -------------------------------------------------------------------------------- /2017年12月/20171212 第71期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171212 第71期/README.md -------------------------------------------------------------------------------- /2017年12月/20171213 第72期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171213 第72期/README.md -------------------------------------------------------------------------------- /2017年12月/20171214 第73期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171214 第73期/README.md -------------------------------------------------------------------------------- /2017年12月/20171215 第74期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171215 第74期/README.md -------------------------------------------------------------------------------- /2017年12月/20171216 第75期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171216 第75期/README.md -------------------------------------------------------------------------------- /2017年12月/20171217 第76期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171217 第76期/README.md -------------------------------------------------------------------------------- /2017年12月/20171218 第77期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171218 第77期/README.md -------------------------------------------------------------------------------- /2017年12月/20171219 第78期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171219 第78期/README.md -------------------------------------------------------------------------------- /2017年12月/20171220 第79期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171220 第79期/README.md -------------------------------------------------------------------------------- /2017年12月/20171221 第80期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171221 第80期/README.md -------------------------------------------------------------------------------- /2017年12月/20171222 第81期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171222 第81期/README.md -------------------------------------------------------------------------------- /2017年12月/20171223 第82期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171223 第82期/README.md -------------------------------------------------------------------------------- /2017年12月/20171224 第83期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171224 第83期/README.md -------------------------------------------------------------------------------- /2017年12月/20171225 第84期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171225 第84期/README.md -------------------------------------------------------------------------------- /2017年12月/20171226 第85期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171226 第85期/README.md -------------------------------------------------------------------------------- /2017年12月/20171227 第86期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171227 第86期/README.md -------------------------------------------------------------------------------- /2017年12月/20171228 第87期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171228 第87期/README.md -------------------------------------------------------------------------------- /2017年12月/20171229 第88期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171229 第88期/README.md -------------------------------------------------------------------------------- /2017年12月/20171230 第89期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171230 第89期/README.md -------------------------------------------------------------------------------- /2017年12月/20171231 第90期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2017年12月/20171231 第90期/README.md -------------------------------------------------------------------------------- /2018年1月/20180101 第91期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180101 第91期/README.md -------------------------------------------------------------------------------- /2018年1月/20180102 第92期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180102 第92期/README.md -------------------------------------------------------------------------------- /2018年1月/20180103 第93期/.md: -------------------------------------------------------------------------------- 1 | ![](https://media.licdn.com/dms/image/C4D12AQG6uImV4n37jg/article-inline_image-shrink_1000_1488/0?e=1560384000&v=beta&t=YaOaA1VZ0RZelhM9k4TiJtzfevgcnqAhiIWy0zLPDUc) 2 | 3 | 4 | 5 | Graphic taken from Data Science Central 6 | 7 | 8 | 9 | The picture is accurate, but the more relevant question is “When would each technique be at an advantage?”   10 | 11 | 12 | 13 | The obvious difference, correctly depicted, is that the Deep Neural Network is estimating many more parameters and even more permutations of parameters than the logistic regression. Therefore the real question is in what situations would that be a good idea?  14 | 15 | 16 | 17 | You need a good ratio of data points to parameters to get reliable estimates so the first criteria would be lots of data in order to estimate lots of parameters. If that's not true then you'd be estimating lots of parameters with little data per parameter and get a bunch of spurious results. Therefore depending upon the situation, the additional granularity of the Deep Neural Network would either represent a treasure trove of additional detail and value, or an error prone and misleading representation of the situation.  18 | 19 | 20 | 21 | The second key difference is the need to understand “why the prediction works” or the need to restrict the equation from using certain data in specific ways. We've all heard the example of drownings and ice cream sales being correlated together because more people both swim and drown in the summer time and also eat more ice cream in the summer. Ice cream sales might help indicate “when people will drown”, but it’s not going to indicate “why people are drowning”. The need to know “why” means that it’s important to restrict the ways data is used and assure logical inference. The more convoluted the formula and the less involved the analyst is the less you’ll be able to understand what caused what or why a prediction works and when it might stop working. Understanding the why is better suited to more parsimonious techniques with careful involvement of the analyst. 22 | 23 | 24 | 25 | On the other hand sometimes the “why” is not as important as simply “what is”. A ground breaking application of Deep Neural Networks is in the area of machine vision or the correct classification of pictures or the translation of video into analyzable data. Pictures and video have enormous amounts of information and detail. So much so that it’s very difficult to effectively use it all without heavy automation. That’s the perfect fit for a Deep Learning Neural Network. 26 | 27 | 28 | 29 | Both techniques, as well as their many cousins, have tremendous opportunities to add value if applied to the problems they’re best suited for and conversely, as with any technique, they could also lead to problems if naively applied inappropriately. 30 | 31 | 32 | 33 | If you liked this discussion, I’d appreciate you sharing it or clicking the “like” button. Your vote of approval helps spread the publicity and is always appreciated and useful in the prioritization of further content. 34 | 35 | 36 | 37 | **David Young has worked in Marketing Analytics 20+ years and lives in Vienna, VA** 38 | 39 | 40 | 41 | If you enjoyed this you might enjoy my book: 42 | 43 | 44 | 45 | Book preview and always in stock at the Publisher: [https://store.bookbaby.com//bookshop/book/index.aspx?bookURL=A-Short-Guide-to-Marketing-Model-Alignment-and-Design](https://store.bookbaby.com/bookshop/book/index.aspx?bookURL=A-Short-Guide-to-Marketing-Model-Alignment-and-Design) 46 | 47 | 48 | 49 | Also at Amazon: [https://www.amazon.com/Short-Guide-Marketing-Alignment-Design/dp/1543912591/ref=sr_1_1?ie=UTF8&qid=1510196791&sr=8-1&keywords=a+short+guide+to+marketing+model+alignment+%26+design](https://www.amazon.com/Short-Guide-Marketing-Alignment-Design/dp/1543912591/ref=sr_1_1?ie=UTF8&qid=1510196791&sr=8-1&keywords=a+short+guide+to+marketing+model+alignment+&+design) 50 | 51 | -------------------------------------------------------------------------------- /2018年1月/20180103 第93期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180103 第93期/README.md -------------------------------------------------------------------------------- /2018年1月/20180103 第93期/logistic-regression-vs-deep-neural-networks-david.md: -------------------------------------------------------------------------------- 1 | ![](https://media.licdn.com/dms/image/C4D12AQG6uImV4n37jg/article-inline_image-shrink_1000_1488/0?e=1560384000&v=beta&t=YaOaA1VZ0RZelhM9k4TiJtzfevgcnqAhiIWy0zLPDUc) 2 | 3 | 4 | 5 | Graphic taken from Data Science Central 6 | 7 | 8 | 9 | The picture is accurate, but the more relevant question is “When would each technique be at an advantage?”   10 | 11 | 12 | 13 | The obvious difference, correctly depicted, is that the Deep Neural Network is estimating many more parameters and even more permutations of parameters than the logistic regression. Therefore the real question is in what situations would that be a good idea?  14 | 15 | 16 | 17 | You need a good ratio of data points to parameters to get reliable estimates so the first criteria would be lots of data in order to estimate lots of parameters. If that's not true then you'd be estimating lots of parameters with little data per parameter and get a bunch of spurious results. Therefore depending upon the situation, the additional granularity of the Deep Neural Network would either represent a treasure trove of additional detail and value, or an error prone and misleading representation of the situation.  18 | 19 | 20 | 21 | The second key difference is the need to understand “why the prediction works” or the need to restrict the equation from using certain data in specific ways. We've all heard the example of drownings and ice cream sales being correlated together because more people both swim and drown in the summer time and also eat more ice cream in the summer. Ice cream sales might help indicate “when people will drown”, but it’s not going to indicate “why people are drowning”. The need to know “why” means that it’s important to restrict the ways data is used and assure logical inference. The more convoluted the formula and the less involved the analyst is the less you’ll be able to understand what caused what or why a prediction works and when it might stop working. Understanding the why is better suited to more parsimonious techniques with careful involvement of the analyst. 22 | 23 | 24 | 25 | On the other hand sometimes the “why” is not as important as simply “what is”. A ground breaking application of Deep Neural Networks is in the area of machine vision or the correct classification of pictures or the translation of video into analyzable data. Pictures and video have enormous amounts of information and detail. So much so that it’s very difficult to effectively use it all without heavy automation. That’s the perfect fit for a Deep Learning Neural Network. 26 | 27 | 28 | 29 | Both techniques, as well as their many cousins, have tremendous opportunities to add value if applied to the problems they’re best suited for and conversely, as with any technique, they could also lead to problems if naively applied inappropriately. 30 | 31 | 32 | 33 | If you liked this discussion, I’d appreciate you sharing it or clicking the “like” button. Your vote of approval helps spread the publicity and is always appreciated and useful in the prioritization of further content. 34 | 35 | 36 | 37 | **David Young has worked in Marketing Analytics 20+ years and lives in Vienna, VA** 38 | 39 | 40 | 41 | If you enjoyed this you might enjoy my book: 42 | 43 | 44 | 45 | Book preview and always in stock at the Publisher: [https://store.bookbaby.com//bookshop/book/index.aspx?bookURL=A-Short-Guide-to-Marketing-Model-Alignment-and-Design](https://store.bookbaby.com/bookshop/book/index.aspx?bookURL=A-Short-Guide-to-Marketing-Model-Alignment-and-Design) 46 | 47 | 48 | 49 | Also at Amazon: [https://www.amazon.com/Short-Guide-Marketing-Alignment-Design/dp/1543912591/ref=sr_1_1?ie=UTF8&qid=1510196791&sr=8-1&keywords=a+short+guide+to+marketing+model+alignment+%26+design](https://www.amazon.com/Short-Guide-Marketing-Alignment-Design/dp/1543912591/ref=sr_1_1?ie=UTF8&qid=1510196791&sr=8-1&keywords=a+short+guide+to+marketing+model+alignment+&+design) 50 | 51 | -------------------------------------------------------------------------------- /2018年1月/20180104 第94期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180104 第94期/README.md -------------------------------------------------------------------------------- /2018年1月/20180105 第95期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180105 第95期/README.md -------------------------------------------------------------------------------- /2018年1月/20180106 第96期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180106 第96期/README.md -------------------------------------------------------------------------------- /2018年1月/20180107 第97期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180107 第97期/README.md -------------------------------------------------------------------------------- /2018年1月/20180108 第98期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180108 第98期/README.md -------------------------------------------------------------------------------- /2018年1月/20180109 第99期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180109 第99期/README.md -------------------------------------------------------------------------------- /2018年1月/20180110 第100期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180110 第100期/README.md -------------------------------------------------------------------------------- /2018年1月/20180111 第101期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180111 第101期/README.md -------------------------------------------------------------------------------- /2018年1月/20180112 第102期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180112 第102期/README.md -------------------------------------------------------------------------------- /2018年1月/20180112 第102期/the-8-neural-network-architectures-machine-learning-researchers-need-to-learn.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180112 第102期/the-8-neural-network-architectures-machine-learning-researchers-need-to-learn.md -------------------------------------------------------------------------------- /2018年1月/20180113 第103期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180113 第103期/README.md -------------------------------------------------------------------------------- /2018年1月/20180114 第104期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180114 第104期/README.md -------------------------------------------------------------------------------- /2018年1月/20180115 第105期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180115 第105期/README.md -------------------------------------------------------------------------------- /2018年1月/20180116 第106期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180116 第106期/README.md -------------------------------------------------------------------------------- /2018年1月/20180117 第107期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180117 第107期/README.md -------------------------------------------------------------------------------- /2018年1月/20180118 第108期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180118 第108期/README.md -------------------------------------------------------------------------------- /2018年1月/20180119 第109期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180119 第109期/README.md -------------------------------------------------------------------------------- /2018年1月/20180120 第110期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180120 第110期/README.md -------------------------------------------------------------------------------- /2018年1月/20180121 第111期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180121 第111期/README.md -------------------------------------------------------------------------------- /2018年1月/20180122 第112期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180122 第112期/README.md -------------------------------------------------------------------------------- /2018年1月/20180123 第113期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180123 第113期/README.md -------------------------------------------------------------------------------- /2018年1月/20180124 第114期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180124 第114期/README.md -------------------------------------------------------------------------------- /2018年1月/20180125 第115期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180125 第115期/README.md -------------------------------------------------------------------------------- /2018年1月/20180126 第116期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180126 第116期/README.md -------------------------------------------------------------------------------- /2018年1月/20180127 第117期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180127 第117期/README.md -------------------------------------------------------------------------------- /2018年1月/20180128 第118期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180128 第118期/README.md -------------------------------------------------------------------------------- /2018年1月/20180129 第119期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180129 第119期/README.md -------------------------------------------------------------------------------- /2018年1月/20180130 第120期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180130 第120期/README.md -------------------------------------------------------------------------------- /2018年1月/20180131 第121期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年1月/20180131 第121期/README.md -------------------------------------------------------------------------------- /2018年2月/20180201 第122期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180201 第122期/README.md -------------------------------------------------------------------------------- /2018年2月/20180202 第123期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180202 第123期/README.md -------------------------------------------------------------------------------- /2018年2月/20180203 第124期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180203 第124期/README.md -------------------------------------------------------------------------------- /2018年2月/20180204 第125期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180204 第125期/README.md -------------------------------------------------------------------------------- /2018年2月/20180205 第126期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180205 第126期/README.md -------------------------------------------------------------------------------- /2018年2月/20180206 第127期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180206 第127期/README.md -------------------------------------------------------------------------------- /2018年2月/20180207 第128期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180207 第128期/README.md -------------------------------------------------------------------------------- /2018年2月/20180208 第129期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180208 第129期/README.md -------------------------------------------------------------------------------- /2018年2月/20180209 第130期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180209 第130期/README.md -------------------------------------------------------------------------------- /2018年2月/20180210 第131期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180210 第131期/README.md -------------------------------------------------------------------------------- /2018年2月/20180211 第132期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180211 第132期/README.md -------------------------------------------------------------------------------- /2018年2月/20180212 第133期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180212 第133期/README.md -------------------------------------------------------------------------------- /2018年2月/20180213 第134期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180213 第134期/README.md -------------------------------------------------------------------------------- /2018年2月/20180214 第135期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180214 第135期/README.md -------------------------------------------------------------------------------- /2018年2月/20180215 第136期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180215 第136期/README.md -------------------------------------------------------------------------------- /2018年2月/20180216 第137期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180216 第137期/README.md -------------------------------------------------------------------------------- /2018年2月/20180217 第138期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180217 第138期/README.md -------------------------------------------------------------------------------- /2018年2月/20180218 第139期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180218 第139期/README.md -------------------------------------------------------------------------------- /2018年2月/20180219 第140期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180219 第140期/README.md -------------------------------------------------------------------------------- /2018年2月/20180219 第140期/mltest-automatically-test-neural-network-models-in-one-function-call.md: -------------------------------------------------------------------------------- 1 | # mltest: Automatically test neural network models in one function call 2 | 3 | ![](https://cdn-images-1.medium.com/max/1600/1*nwfCyEi6d3Jt2CfcC-2_WA.png) 4 | 5 | So I got a lot of positive feedback on my last post on [how to unit test machine learning code](https://medium.com/@keeper6928/how-to-unit-test-machine-learning-code-57cf6fd81765). A few people actually messaged me directly saying they caught a bug in their own code with the recommended tests, which is awesome! But these issues are still too common, and it is just as easy to forget to write a test as it is to write the bug in the first place. We need a better, more automated solution. 6 | 7 | That is why we are introducing [mltest: Automated ML testing in one function call.](https://github.com/Thenerdstation/mltest) 8 | 9 | Check it out! 10 | 11 | 12 | 13 | Done. With incredibly little setup, we now are testing against several different common machine learning issues. 14 | 15 | To install it, just run: 16 | 17 | 18 | 19 | The function call mltest.test_suite(…) is the main powerhouse of this library. It runs several tests including: 20 | 21 | ### 1. Variables change 22 | 23 | The test from [my first post](https://medium.com/@keeper6928/how-to-unit-test-machine-learning-code-57cf6fd81765) that helped people the most was the variables change test. Basically, you run the train_op and make sure that all of the variables within a list or scope are modified. 24 | 25 | ### 2. Variables DON’T change 26 | 27 | It is also possible to make sure that only variables within a scope or a list are the ones that change, and that the rest do not change. This is super useful for GAN training, as the generator and discriminator usually have different training ops. 28 | 29 | ### 3. Logits Range 30 | 31 | One common mistake that I would make is adding a non-linearity to my logits output. This would usually cause lots of problems once it hits softmax. Our test_suite() automatically checks that the logits layer has values that are above and below 0. This perhaps isn’t the best way of detecting it, but it helps narrow down the problem. 32 | 33 | You can also set the range to check to be whatever you want. Let’s say you expect to have a tanh on the logits, you can set a check to make sure all of the values of the logits are in (-1, 1). 34 | 35 | ### 4. Input dependencies 36 | 37 | One common issue is that sometimes people forget to connect the branches of their network together. Whether you forget to add two tensors together or forget to call the function call for a certain branch. A network can still train and converge poorly with only partial input, so it is import to make sure all of your input values are dependents of the training op. 38 | 39 | Of course, any of these tests can be turned off manually with flags in test_suite(). See the code for documentation on how to do this. 40 | 41 | ### mltest setup 42 | 43 | Another useful feature is mltest.setup(). 44 | 45 | 46 | 47 | This call will automatically reset the default tensorflow graph and set tensorflow’s, numpy’s, and python’s random seeds. It’s very easy to forget to seed your random values, and can cause a massive headache when trying to recreate bugs. 48 | 49 | This suite is still in beta, so if you have any requests or notice any bugs, please add an issue on Github! Also, I accept pull requests. ;) 50 | 51 | -------------------------------------------------------------------------------- /2018年2月/20180220 第141期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180220 第141期/README.md -------------------------------------------------------------------------------- /2018年2月/20180221 第142期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180221 第142期/README.md -------------------------------------------------------------------------------- /2018年2月/20180222 第143期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180222 第143期/README.md -------------------------------------------------------------------------------- /2018年2月/20180223 第144期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180223 第144期/README.md -------------------------------------------------------------------------------- /2018年2月/20180224 第145期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180224 第145期/README.md -------------------------------------------------------------------------------- /2018年2月/20180225 第146期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180225 第146期/README.md -------------------------------------------------------------------------------- /2018年2月/20180226 第147期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180226 第147期/README.md -------------------------------------------------------------------------------- /2018年2月/20180227 第148期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180227 第148期/README.md -------------------------------------------------------------------------------- /2018年2月/20180228 第149期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年2月/20180228 第149期/README.md -------------------------------------------------------------------------------- /2018年3月/20180301 第150期/README.md: -------------------------------------------------------------------------------- 1 | | 标题 | 简介 | 2 | | --- | --- | 3 | | [实时人脸识别：一个端到端的项目](https://www.hackster.io/mjrobot/real-time-face-recognition-an-end-to-end-project-a10826?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 4 | | [101个用于数据分析的NumPy练习](https://www.machinelearningplus.com/python/101-numpy-exercises-python/) | | 5 | | [深度解密换脸应用Deepfake](https://zhuanlan.zhihu.com/p/34042498?from=hackcv&hmsr=hackcv.com&utm_medium=hackcv.com&utm_source=hackcv.com) | | 6 | -------------------------------------------------------------------------------- /2018年3月/20180302 第151期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180302 第151期/README.md -------------------------------------------------------------------------------- /2018年3月/20180303 第152期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180303 第152期/README.md -------------------------------------------------------------------------------- /2018年3月/20180304 第153期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180304 第153期/README.md -------------------------------------------------------------------------------- /2018年3月/20180305 第154期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180305 第154期/README.md -------------------------------------------------------------------------------- /2018年3月/20180306 第155期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180306 第155期/README.md -------------------------------------------------------------------------------- /2018年3月/20180307 第156期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180307 第156期/README.md -------------------------------------------------------------------------------- /2018年3月/20180308 第157期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180308 第157期/README.md -------------------------------------------------------------------------------- /2018年3月/20180309 第158期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180309 第158期/README.md -------------------------------------------------------------------------------- /2018年3月/20180310 第159期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180310 第159期/README.md -------------------------------------------------------------------------------- /2018年3月/20180311 第160期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180311 第160期/README.md -------------------------------------------------------------------------------- /2018年3月/20180312 第161期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180312 第161期/README.md -------------------------------------------------------------------------------- /2018年3月/20180313 第162期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180313 第162期/README.md -------------------------------------------------------------------------------- /2018年3月/20180314 第163期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180314 第163期/README.md -------------------------------------------------------------------------------- /2018年3月/20180315 第164期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180315 第164期/README.md -------------------------------------------------------------------------------- /2018年3月/20180316 第165期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180316 第165期/README.md -------------------------------------------------------------------------------- /2018年3月/20180317 第166期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180317 第166期/README.md -------------------------------------------------------------------------------- /2018年3月/20180318 第167期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180318 第167期/README.md -------------------------------------------------------------------------------- /2018年3月/20180319 第168期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180319 第168期/README.md -------------------------------------------------------------------------------- /2018年3月/20180320 第169期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180320 第169期/README.md -------------------------------------------------------------------------------- /2018年3月/20180321 第170期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180321 第170期/README.md -------------------------------------------------------------------------------- /2018年3月/20180322 第171期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180322 第171期/README.md -------------------------------------------------------------------------------- /2018年3月/20180323 第172期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180323 第172期/README.md -------------------------------------------------------------------------------- /2018年3月/20180324 第173期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180324 第173期/README.md -------------------------------------------------------------------------------- /2018年3月/20180325 第174期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180325 第174期/README.md -------------------------------------------------------------------------------- /2018年3月/20180326 第175期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180326 第175期/README.md -------------------------------------------------------------------------------- /2018年3月/20180327 第176期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180327 第176期/README.md -------------------------------------------------------------------------------- /2018年3月/20180328 第177期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180328 第177期/README.md -------------------------------------------------------------------------------- /2018年3月/20180329 第178期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180329 第178期/README.md -------------------------------------------------------------------------------- /2018年3月/20180330 第179期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180330 第179期/README.md -------------------------------------------------------------------------------- /2018年3月/20180331 第180期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年3月/20180331 第180期/README.md -------------------------------------------------------------------------------- /2018年4月/20180401 第181期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180401 第181期/README.md -------------------------------------------------------------------------------- /2018年4月/20180402 第182期/50-must-read-books-for-machine-learning.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180402 第182期/50-must-read-books-for-machine-learning.md -------------------------------------------------------------------------------- /2018年4月/20180402 第182期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180402 第182期/README.md -------------------------------------------------------------------------------- /2018年4月/20180403 第183期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180403 第183期/README.md -------------------------------------------------------------------------------- /2018年4月/20180404 第184期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180404 第184期/README.md -------------------------------------------------------------------------------- /2018年4月/20180405 第185期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180405 第185期/README.md -------------------------------------------------------------------------------- /2018年4月/20180406 第186期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180406 第186期/README.md -------------------------------------------------------------------------------- /2018年4月/20180407 第187期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180407 第187期/README.md -------------------------------------------------------------------------------- /2018年4月/20180408 第188期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180408 第188期/README.md -------------------------------------------------------------------------------- /2018年4月/20180409 第189期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180409 第189期/README.md -------------------------------------------------------------------------------- /2018年4月/20180410 第190期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180410 第190期/README.md -------------------------------------------------------------------------------- /2018年4月/20180411 第191期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180411 第191期/README.md -------------------------------------------------------------------------------- /2018年4月/20180412 第192期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180412 第192期/README.md -------------------------------------------------------------------------------- /2018年4月/20180413 第193期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180413 第193期/README.md -------------------------------------------------------------------------------- /2018年4月/20180413 第193期/sentiment-classification-from-keras-to-the-browser.md: -------------------------------------------------------------------------------- 1 | # Sentiment Classification from Keras to the Browser 2 | 3 | We will create a classifier using Keras to differentiate between positive sentiments and negative sentiments for movies. Then we will load the model to the browser for user prediction. 4 | 5 | #### The Code 6 | 7 | I created a repository on [GitHub](https://github.com/zaidalyafeai/Browser-Sentiment-Classification) with the code required to follow the tutorial. If you spot any error or faced any problems please raise an issue there. 8 | 9 | #### The Dataset 10 | 11 | We will use the [dataset](https://www.kaggle.com/c/si650winter11/data) for sentiment classification. The dataset contains `7086` statements about movies with labels. label `1` means positive sentiment and `0` means negative sentiment. 12 | 13 | #### Preprocessing the data 14 | 15 | For text analysis we need first to preprocess the data. First of all, upon uploading the data we need to split the sentences into words then populate the training set. 16 | 17 | 18 | 19 | Then after that we will need to create the dictionary `word_index` that will map words to integers for embedding. Note that the dictionary length might be huge so we count the counts of each word in the training set and we only include words that repeat more than some threshold. 20 | 21 | 22 | 23 | The process method will create words out of statements by removing punctuations 24 | 25 | 26 | 27 | After that we will need to map each sentimental statement to a sequence of integers but first get the statement with the largest number of words 28 | 29 | 30 | 31 | Now we can create a sequence of integers using our dictionary `word_index` Note that we create a fixed number of sequences with length `max_tokens` which has the largest number of tokens. We use pre padding since it gives better results 32 | 33 | 34 | 35 | Now we are ready to create the keras model. The first layer is the embedding layer. Then we follow it with 3 [GRU](https://en.wikipedia.org/wiki/Gated_recurrent_unit) layers. Finally we create a dense layer with sigmoid activation. We compile the model using an Adam optimizer 36 | 37 | 38 | 39 | Then we train the model for `5` epochs with `5%` split for validation with `32` batch size 40 | 41 | 42 | 43 | We reach `97%` accuracy on validation set. 44 | 45 | ![](https://cdn-images-1.medium.com/max/1600/1*V0l012DyGMDERgdHt7fKWw.png) 46 | 47 | We can see the model summary using 48 | 49 | 50 | 51 | ![](https://cdn-images-1.medium.com/max/1600/1*e3lIBA3Cz33FQYiLrOQ89Q.png) 52 | 53 | We will save the model using `.h5` format 54 | 55 | 56 | 57 | Now that we are done with the model we will import it to run on the browser. First we will need to convert it into json format. Before this step you will need to install tensorflowjs tools using 58 | 59 | 60 | 61 | Convert the keras model into a model understood by tensorflowjs 62 | 63 | 64 | 65 | This will create one json file which contains the meta variables and some other variables with names like `group1-shard1of1` which contains the computed values of the weights 66 | 67 | #### Porting the model to the browser 68 | 69 | For simple processing we will load the dictionary generated by the keras code 70 | 71 | 72 | 73 | Hence `word_index` will now contain the same word,index pair as in the previous section. After that we need to create some helper methods to process the text, `tokenize` it then map it to integers using the dictionary 74 | 75 | 76 | 77 | Now we can load that into the browser using `tensorflow.js` 78 | 79 | 80 | 81 | Finally we combine all the steps using one method 82 | 83 | 84 | 85 | Using the sample text we get a prediction value `0.98 `which is very close to `1`. 86 | 87 | #### References 88 | 89 | [**Hvass-Labs/TensorFlow-Tutorials** 90 | TensorFlow-Tutorials - TensorFlow Tutorials with YouTube Videosgithub.com](https://github.com/Hvass-Labs/TensorFlow-Tutorials)[](https://github.com/Hvass-Labs/TensorFlow-Tutorials) 91 | 92 | [**tensorflow/tfjs** 93 | tfjs - A WebGL accelerated, browser based JavaScript library for training and deploying ML models.github.com](https://github.com/tensorflow/tfjs)[](https://github.com/tensorflow/tfjs) 94 | 95 | -------------------------------------------------------------------------------- /2018年4月/20180414 第194期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180414 第194期/README.md -------------------------------------------------------------------------------- /2018年4月/20180415 第195期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180415 第195期/README.md -------------------------------------------------------------------------------- /2018年4月/20180416 第196期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180416 第196期/README.md -------------------------------------------------------------------------------- /2018年4月/20180417 第197期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180417 第197期/README.md -------------------------------------------------------------------------------- /2018年4月/20180418 第198期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180418 第198期/README.md -------------------------------------------------------------------------------- /2018年4月/20180419 第199期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180419 第199期/README.md -------------------------------------------------------------------------------- /2018年4月/20180420 第200期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180420 第200期/README.md -------------------------------------------------------------------------------- /2018年4月/20180421 第201期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180421 第201期/README.md -------------------------------------------------------------------------------- /2018年4月/20180422 第202期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180422 第202期/README.md -------------------------------------------------------------------------------- /2018年4月/20180423 第203期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180423 第203期/README.md -------------------------------------------------------------------------------- /2018年4月/20180424 第204期/5-ways-of-looking-at-machine-learning-what-is-ml.md: -------------------------------------------------------------------------------- 1 | # 5 ways of looking at Machine Learning… What is ML? 2 | 3 | **1)** 4 | 5 | ![](https://cdn-images-1.medium.com/max/1600/1*ZWKCLqcegxPYSTUNJ-92Ag.png) 6 | 7 | **2)** We have solved lot of 2 equation 2 unknown problem. 8 | 9 | 2x + 3y =5 10 | 11 | -4x+7y = 9 12 | 13 | Solving Machine learning problem means exactly solving these type of equation. only difference is that in case of ML we have millions of equation, billions of unknown and trillion of possible solution. Our task is to find best solution out of such huge possibility of solution. 14 | 15 | **3) Look at below series :** 16 | 17 | 6,6,6,6,6,6,6,6 18 | 19 | 6,7,6,7,6,7,6,7 20 | 21 | 6,7,8,9,10,11,12,13 22 | 23 | For computer above all are numbers, For us 24 | 25 | in 1st — All are **same** 26 | 27 | in 2nd —**increase and decrease** sequence 28 | 29 | in 3rd — continuously **increasing** 30 | 31 | So, finding such **rule, pattern from Data** is Machine Learning. 32 | 33 | **4)**Automation is all about do work in loop like **for and while loop** of programming language. 34 | 35 | Then Machine Learning is about automatic **building if/else** part of system. 36 | 37 | **5)** Industrial revolution — **Automation** 38 | 39 | Current era, Machine Learning — **Automation of automation** 40 | 41 | Future, AutoML — **Automation of automation of automation** 42 | 43 | Checkout my data science course at just $9.99 44 | 45 | [https://www.udemy.com/data-science-with-python-and-pandas/?couponCode=KNOWLEDGE_IS_POWER](https://www.udemy.com/data-science-with-python-and-pandas/?couponCode=KNOWLEDGE_IS_POWER) 46 | 47 | -------------------------------------------------------------------------------- /2018年4月/20180424 第204期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180424 第204期/README.md -------------------------------------------------------------------------------- /2018年4月/20180425 第205期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180425 第205期/README.md -------------------------------------------------------------------------------- /2018年4月/20180426 第206期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180426 第206期/README.md -------------------------------------------------------------------------------- /2018年4月/20180427 第207期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180427 第207期/README.md -------------------------------------------------------------------------------- /2018年4月/20180428 第208期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180428 第208期/README.md -------------------------------------------------------------------------------- /2018年4月/20180429 第209期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180429 第209期/README.md -------------------------------------------------------------------------------- /2018年4月/20180429 第209期/transfer-learning-will-radically-change-machine-learning-for-engineers.md: -------------------------------------------------------------------------------- 1 | ![](https://cdn-images-1.medium.com/max/1600/1*mmnxXocQ36q3mqYgyxKz7g.jpeg) 2 | 3 | # Transfer Learning Will Radically Change Machine Learning for Engineers 4 | 5 | 6 | 7 | In traditional supervised machine learning, we teach a model to become more successful and efficient at a task, by providing it with example data. Generally, once the model begins to perform well on the training data for the domain or problem it is tasked with, we expect a reasonable performance for new data. But, if you think about it, there are a few issues with this traditional supervised learning process 8 | 9 | As engineers, we’re forced to construct specific models that only excel at a specific problem. This costs us valuable engineering time to create, train and tune models from scratch for every new problem we want to tackle, even if it’s a problem that has been solved in industry. 10 | 11 | From a product perspective, this is incredibly detrimental to progress, and could really hamper feature releases and engineering productivity. Transfer Learning offers an interesting solution to this problem. 12 | 13 | Transfer learning as a paradigm can solve this problem, by allowing us to leverage existing knowledge and data from a certain related domain to the new one we’re trying to train for. In 2016, Andrew Ng posited that Transfer Learning would be essential to commercial and industry success. 14 | 15 | 16 | 17 | #### Applications of Transfer Learning 18 | 19 | Making use of pre-trained models and related domain data promises to supercharge most general development for machine learning. By tapping into a pre-trained model for a related purpose to its original design, your team can leapfrog the data cleaning, setup and training required to bring a model up to par for the task. 20 | 21 | Two common areas where Transfer Learning has already been demonstrated to great success are Images and Text . 22 | 23 | Transfer Learning has been particularly effective with Image Data, and it is common to leverage a deep learning model trained on some large image data set, like ImageNet. These pretrained models can be directly included in other new models that expect some form of Image Input. 24 | 25 | ![](https://cdn-images-1.medium.com/max/1600/1*WyTv5S6b8fmOG6aBrZyiHw.jpeg) 26 | 27 | With Textual data, words are mapped to vectors where different words with a similar meaning have a similar vector representation. Pretrained models exist to learn these representations, and are widely available. These can then be incorporated into deep learning language models, both at the input or output stage. 28 | 29 | ![](https://cdn-images-1.medium.com/max/1600/1*YjO-agjV5Q5nv7BaAGzBsQ.png) 30 | 31 | **Transfer Learning and pretrained models are the future of machine learning applications in general development, and as such need to be made more accessible and discoverable for everyone.** 32 | 33 | **That’s why we’re building**[ModelDepot](https://modeldepot.io)**, to decrease the friction associated with model access and contribute to the democratization of AI in the 21st century.** 34 | 35 | **Join the conversation on**[Gitter](https://gitter.im/ModelDepot/Lobby)**!**👋 36 | 37 | -------------------------------------------------------------------------------- /2018年4月/20180430 第210期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年4月/20180430 第210期/README.md -------------------------------------------------------------------------------- /2018年5月/20180501 第211期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180501 第211期/README.md -------------------------------------------------------------------------------- /2018年5月/20180502 第212期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180502 第212期/README.md -------------------------------------------------------------------------------- /2018年5月/20180503 第213期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180503 第213期/README.md -------------------------------------------------------------------------------- /2018年5月/20180504 第214期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180504 第214期/README.md -------------------------------------------------------------------------------- /2018年5月/20180505 第215期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180505 第215期/README.md -------------------------------------------------------------------------------- /2018年5月/20180505 第215期/colab-an-easy-way-to-learn-and-use-tensorflow.md: -------------------------------------------------------------------------------- 1 | # Colab: An easy way to learn and use TensorFlow 2 | 3 | ![](https://cdn-images-1.medium.com/max/1600/1*g_x1-5iYRn-SmdVucceiWw.png) 4 | 5 | Colaboratory is a hosted Jupyter notebook environment that is free to use and requires no setup. You may have already seen it in [Machine Learning Crash Course](https://developers.google.com/machine-learning/crash-course/), tensorflow.org’s [eager execution](https://colab.research.google.com/github/tensorflow/models/blob/master/samples/core/get_started/eager.ipynb) tutorial, or on various research articles (like [this one](https://distill.pub/2018/building-blocks/)). We wanted to offer 5 tips for using it: 6 | 7 | **1. TensorFlow is already pre-installed** 8 | 9 | When you create a new notebook on [colab.research.google.com](http://colab.research.google.com/), TensorFlow is already pre-installed and optimized for the hardware being used. Just `import tensorflow as tf`, and start coding. 10 | 11 | **2. Setup your libraries and data dependencies in code cells** 12 | 13 | Creating a cell with `!pip install` or `!apt-get` works as you’d expect. It also makes it easy for others to reproduce your setup. 14 | 15 | To get in your training data, you can follow these tutorials for popular data sources: [BigQuery](https://colab.research.google.com/notebooks/bigquery.ipynb), [Drive, Sheets, or Google Cloud Storage](https://colab.research.google.com/notebooks/io.ipynb). You also have access to the shell with `!`, so `!wget`, `!pwd`, etc. might also help. 16 | 17 | **3. Use it with Github** 18 | 19 | If you have a nice .ipynb on Github, it’s easy to create a one-click link for your readers to start playing with it. Just add your Github path to colab.research.google.com/github/ . For example, [colab.research.google.com/github/tensorflow/tensor2tensor/blob/master/tensor2tensor/notebooks/hello_t2t.ipynb](https://colab.research.google.com/github/tensorflow/tensor2tensor/blob/master/tensor2tensor/notebooks/hello_t2t.ipynb) will load [this ipynb](https://github.com/tensorflow/tensor2tensor/blob/master/tensor2tensor/notebooks/hello_t2t.ipynb) stored on Github. 20 | 21 | ![](https://cdn-images-1.medium.com/max/1600/1*ZpNn76K98snC9vDiIJ6Ldw.jpeg) 22 | 23 | You can also easily save a copy of your Colab notebook to Github by using File > Save a copy to Github… 24 | 25 | **4. Share and edit collaboratively** 26 | 27 | Colab notebooks are just like Google Docs and Sheets. They are stored in Google Drive and can be shared, edited, and commented on collaboratively. Just click the Share button in the top right of any notebook that you’ve created. 28 | 29 | **5. Hardware acceleration** 30 | 31 | By default, Colab notebooks run on CPU. You can switch your notebook to run with GPU by going to Runtime > Change runtime type, and then selecting GPU. You can also have a Colab notebook use your local machine’s hardware by following these [instructions](https://research.google.com/colaboratory/local-runtimes.html). 32 | 33 | For more tips, see our [welcome notebook](https://colab.research.google.com/notebooks/welcome.ipynb), read our [FAQ](https://research.google.com/colaboratory/faq.html), or find useful code snippets while using Colab (Help > Search code snippets..). 34 | 35 | Thanks, and we hope you enjoy using TensorFlow and Colab! 36 | 37 | -------------------------------------------------------------------------------- /2018年5月/20180506 第216期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180506 第216期/README.md -------------------------------------------------------------------------------- /2018年5月/20180507 第217期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180507 第217期/README.md -------------------------------------------------------------------------------- /2018年5月/20180508 第218期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180508 第218期/README.md -------------------------------------------------------------------------------- /2018年5月/20180509 第219期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180509 第219期/README.md -------------------------------------------------------------------------------- /2018年5月/20180510 第220期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180510 第220期/README.md -------------------------------------------------------------------------------- /2018年5月/20180511 第221期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180511 第221期/README.md -------------------------------------------------------------------------------- /2018年5月/20180512 第222期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180512 第222期/README.md -------------------------------------------------------------------------------- /2018年5月/20180513 第223期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180513 第223期/README.md -------------------------------------------------------------------------------- /2018年5月/20180514 第224期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180514 第224期/README.md -------------------------------------------------------------------------------- /2018年5月/20180515 第225期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180515 第225期/README.md -------------------------------------------------------------------------------- /2018年5月/20180516 第226期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180516 第226期/README.md -------------------------------------------------------------------------------- /2018年5月/20180517 第227期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180517 第227期/README.md -------------------------------------------------------------------------------- /2018年5月/20180518 第228期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180518 第228期/README.md -------------------------------------------------------------------------------- /2018年5月/20180519 第229期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180519 第229期/README.md -------------------------------------------------------------------------------- /2018年5月/20180520 第230期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180520 第230期/README.md -------------------------------------------------------------------------------- /2018年5月/20180522 第231期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180522 第231期/README.md -------------------------------------------------------------------------------- /2018年5月/20180522 第232期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180522 第232期/README.md -------------------------------------------------------------------------------- /2018年5月/20180523 第233期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180523 第233期/README.md -------------------------------------------------------------------------------- /2018年5月/20180524 第234期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180524 第234期/README.md -------------------------------------------------------------------------------- /2018年5月/20180525 第235期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180525 第235期/README.md -------------------------------------------------------------------------------- /2018年5月/20180526 第236期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180526 第236期/README.md -------------------------------------------------------------------------------- /2018年5月/20180527 第237期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180527 第237期/README.md -------------------------------------------------------------------------------- /2018年5月/20180528 第238期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180528 第238期/README.md -------------------------------------------------------------------------------- /2018年5月/20180529 第239期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180529 第239期/README.md -------------------------------------------------------------------------------- /2018年5月/20180530 第240期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180530 第240期/README.md -------------------------------------------------------------------------------- /2018年5月/20180531 第241期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年5月/20180531 第241期/README.md -------------------------------------------------------------------------------- /2018年6月/20180601 第242期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180601 第242期/README.md -------------------------------------------------------------------------------- /2018年6月/20180602 第243期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180602 第243期/README.md -------------------------------------------------------------------------------- /2018年6月/20180603 第244期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180603 第244期/README.md -------------------------------------------------------------------------------- /2018年6月/20180605 第245期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180605 第245期/README.md -------------------------------------------------------------------------------- /2018年6月/20180605 第246期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180605 第246期/README.md -------------------------------------------------------------------------------- /2018年6月/20180606 第247期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180606 第247期/README.md -------------------------------------------------------------------------------- /2018年6月/20180607 第248期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180607 第248期/README.md -------------------------------------------------------------------------------- /2018年6月/20180608 第249期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180608 第249期/README.md -------------------------------------------------------------------------------- /2018年6月/20180609 第250期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180609 第250期/README.md -------------------------------------------------------------------------------- /2018年6月/20180610 第251期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180610 第251期/README.md -------------------------------------------------------------------------------- /2018年6月/20180611 第252期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180611 第252期/README.md -------------------------------------------------------------------------------- /2018年6月/20180612 第253期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180612 第253期/README.md -------------------------------------------------------------------------------- /2018年6月/20180613 第254期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180613 第254期/README.md -------------------------------------------------------------------------------- /2018年6月/20180614 第255期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180614 第255期/README.md -------------------------------------------------------------------------------- /2018年6月/20180615 第256期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180615 第256期/README.md -------------------------------------------------------------------------------- /2018年6月/20180616 第257期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180616 第257期/README.md -------------------------------------------------------------------------------- /2018年6月/20180617 第258期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180617 第258期/README.md -------------------------------------------------------------------------------- /2018年6月/20180617 第258期/how-to-build-a-neural-network-framework-like-tensorflow-in-c-part-1.md: -------------------------------------------------------------------------------- 1 | # How to Build A Neural Network Framework Like Tensorflow in C++: Part 1 2 | 3 | Alright, so as I discussed in [part 0](https://medium.com/p/56f54c672852/), In this set of tutorials, I’ll be showing how you can build a Neural Network Framework similar to Tensorflow, but in C++. 4 | 5 | The tutorials will be divided into coding a minimum viable product with: 6 | 7 | 1. The cost function 8 | 9 | 2. The minimizer that will minimize the cost function 10 | 11 | 3. The Neural Network module and its back propagation 12 | 13 | In this tutorial, we will talk about the cost function. 14 | 15 | First, let’s see the entire class’s header as it is now, and explain it line by line: 16 | 17 | 18 | 19 | Lot’s of stuff to cover, so let’s get started. 20 | 21 | Lines 1–19 are just the regular includes in a library like Shogun. It includes Stan’s headers(that we discussed in Part 0 of the tutorials) as well as Eigen (which is a linear algebra library in C++) 22 | 23 | Lines 21–28 define some types to make it easier to reference later on. For example, StanVector is an Eigen Matrix that has a bunch of stan variables inside them (See Part 0 for more details). 24 | 25 | Then, we have the main class: StanFirstOrderSAGCostFunction, which provides an interface for defining a stochastic average gradient cost function. It also provides the function get_gradient as well as get_average_gradient which is where most of the work goes. 26 | 27 | But before I explain how this works, we need to look at the members of StanFirstOrderSAGCostFunction. 28 | 29 | The first member is m_X, m_y which are basically the training data and labels of the cost function. 30 | 31 | m_trainable_parameters are as the name suggests a bunch of stan variables that are the parameters of the cost function. 32 | 33 | m_cost_for_ith_point and m_total_cost are again as the name suggests functions which evaluate the error of the ith point with respect to the trainable parameters, as well as the total cost with respect to all the costs of the ith datapoints. 34 | 35 | So, what happens when get_gradient() is called is that we evaluate the errors with respect to the current parameters using the definitions of m_total_cost, then use stan to get the gradient of this error cost function with respect to each of the trainable parameters, and that’s the power of stan! For implementation details, check out this link where I implemented the class: 36 | 37 | 38 | 39 | With this class done, we can now define any arbitrary cost function in terms of stan, and calculate the gradient of it with respect to all parameters using stan. For an example of how to use the class, checkout an example of it being used here, where I implemented mean squared error: 40 | 41 | 42 | 43 | ### A Word on Shogun 44 | 45 | In case you’re interested in joining Open Source, Shogun is a great place to start. They have a super supportive community, and they welcome new comers to Opensource, so swing by and see if you can help with some of the issues labelled “beginner friendly” on their github! 46 | 47 | -------------------------------------------------------------------------------- /2018年6月/20180618 第259期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180618 第259期/README.md -------------------------------------------------------------------------------- /2018年6月/20180619 第260期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180619 第260期/README.md -------------------------------------------------------------------------------- /2018年6月/20180620 第261期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180620 第261期/README.md -------------------------------------------------------------------------------- /2018年6月/20180620 第261期/ncrf-deep-learning-framework-improves-cancer-metastasis-detection.md: -------------------------------------------------------------------------------- 1 | # NCRF Deep Learning Framework Improves Cancer Metastasis Detection 2 | 3 | ![](https://cdn-images-1.medium.com/max/1600/1*Plc8FcIEcyKKkSK88HF9LA.png) 4 | 5 | The **Baidu Silicon Valley Artificial Intelligence Lab** has released a paper which proposes a neural conditional random field (NCRF) process for cancer metastasis detection on Whole Slide Images (WSIs). [Cancer Metastasis Detection With Neural Conditional Random Field. Medical Imaging with Deep Learning (MIDL), 2018, Yi Li and Wei Ping](https://openreview.net/forum?id=S1aY66iiM) has been accepted by the International Conference on Medical Imaging with Deep Learning (MIDL), which runs July 4 ‑ 6 in Amsterdam. 6 |   7 | WSIs are outrageously large files, which usually contain billions of pixels and take up gigabytes of disk space. Pathologists must search these images for groups of tumor cells which can be smaller than 1000 pixels in diameter. Accordingly, to effectively detect metastasis in these massive files is complicated and time-consuming for human experts. 8 | 9 | ![](https://cdn-images-1.medium.com/max/1600/1*k_uQsjz3tVw89FyhdN5olQ.png) 10 | 11 | Various deep learning based algorithms have been proposed to aid pathologists in effectively reviewing these slides. Most of the algorithms split the slide into smaller individual image patches, e.g. at 256x256 pixels. A deep convolutional neural network (CNN) is then trained to determine whether each small patch contains tumor cells or healthy cells. It can however be difficult to predict whether such a patch contains tumor cells without viewing its surroundings, especially when dealing with tumor/healthy boundary regions, and false positive predictions are often returned. 12 | 13 | ![](https://cdn-images-1.medium.com/max/1600/1*erYvHdYS6MCxDlPiefqE7Q.png) 14 | 15 | The NCRF proposed in the paper addresses this issue by**including a grid of neighboring patches as input to provide context that can improve prediction of tumor cells or healthy cells**. 16 | 17 | ![](https://cdn-images-1.medium.com/max/1600/1*Plc8FcIEcyKKkSK88HF9LA.png) 18 | 19 | Conditional random fields (CRF) are used to model the spatial correlations between neighboring patches. The entire NCRF can be trained end-to-end without any pre- or post-processing. 20 |   21 | The major improvement this algorithm provides is that it returns far fewer false positives. The model achieved an average FROC score (a tumor localization rating) of 0.8096 on the Camelyon16 dataset, outperforming not only a professional pathologist (0.7240), but also the previous AI winner of the Camelyon16 challenge (0.8074).  22 |   23 | Further clinical study on larger datasets will be necessary to assess the proposed algorithm comprehensively. 24 |   25 | With the help of better tumor detection algorithms, pathologists can be freed from searching through an entire slide and can focus more on tumor regions highlighted by the algorithm. 26 |   27 |  NCRF is open sourced at GitHub ([https://github.com/baidu-research/NCRF](https://github.com/baidu-research/NCRF)). 28 | 29 | **Author:** Mos Zhang | **Editor:** Michael Sarazen 30 | 31 | **Follow us on Twitter**[@Synced_Global](https://twitter.com/Synced_Global)**for more AI updates!** 32 | 33 | **Subscribe**[here ](https://t.co/d2OrjqTGDq)**to get insightful tech news, reviews and analysis!** 34 | 35 | Let’s talk about AI and FinTech! Synced invites you to join our**DTalk Episode One: Deploying AI in Mobile-First Customer-facing Financial Products: A Tale of Two Cycles**. Jike Chong will share his ideas on employing AI techniques in FinTech business model. Register at [https://goo.gl/KKhHgv](https://t.co/mllmSDfdpU)! We hope to see you on June 21st in Silicon Valley. 36 | 37 | -------------------------------------------------------------------------------- /2018年6月/20180620 第261期/simple-website-text-scraping-with-go-and-aws-lambda.md: -------------------------------------------------------------------------------- 1 | # Simple Website Text Scraping with Go and AWS Lambda 2 | 3 | ![](https://cdn-images-1.medium.com/max/1600/1*bVjCHDdxSvAJrit52W_SoQ.jpeg) 4 | 5 | Recently I needed to know when certain websites were updated with specific text. I decided to utilize AWS Lambda to save on cost of hosting a server, and use Go because it’s fast, and also because it’s one of the supported languages on AWS Lambda. I am also using AWS SES to send me e-mail notifications when results are found. 6 | 7 | Bellow I’ll be showing you how to compile the Go script, setup the AWS Lambda function, and configure a cron type job to run the script every hour. 8 | 9 | First, clone the repo contains the script. 10 | 11 | [https://github.com/aaronvb/aws_lambda_go_scraper](https://github.com/aaronvb/aws_lambda_go_scraper) 12 | 13 | 14 | 15 | Then we’ll build the Go script and zip it up for AWS Lambda. 16 | 17 | 18 | 19 | Create an AWS Lambda function with the Go runtime, and select or create a role that has access to AWS SES. We’ll be using AWS SES to send out the e-mail notification. 20 | 21 | Once the AWS Lambda function is created, upload the zip file and make sure the handler is set to `main`. 22 | 23 | ![](https://cdn-images-1.medium.com/max/1600/1*9ZmhVFoa-6aExvJssU6MaA.png) 24 | 25 | Next, create 3 environment variables: `RECIPIENT` will be the email which receives the notification, `SENDER` which will be the email address that sends the notification, and last `SES_LOCATION` which is the location of your SES(ie: us-west-2). 26 | 27 | ![](https://cdn-images-1.medium.com/max/1600/1*8RWqMXKVrXHhVZ5PUjYTgQ.png) 28 | 29 | Don’t forget to add the email address to SES and verify it so it can receive emails. 30 | 31 | ![](https://cdn-images-1.medium.com/max/1600/1*E304gm1-PjY512hkYSO33w.png) 32 | 33 | Now we can create a test event. In the event data pass a JSON hash which has a key `urls` and a string value with the urls you want to scrape, separated by commas, and a key `words`, with a string value of comma separated words you wish to scrape. 34 | 35 | Example: 36 | 37 | 38 | 39 | ![](https://cdn-images-1.medium.com/max/1600/1*a9KhHh7UMEsx2reSc2wDIA.png) 40 | 41 | ![](https://cdn-images-1.medium.com/max/1600/1*Z27m6X5eDJx_ynwh2g5gbg.png) 42 | 43 | Click the test button and you should receive a successful function execution with logs and an email. The logs will contain the results, message ID from SES, and any errors while parsing or sending the email. 44 | 45 | ![](https://cdn-images-1.medium.com/max/1600/1*PJUQ1PgTMMjn12GCCqQiIQ.png) 46 | 47 | ### Let’s Automate This 48 | 49 | Now that the AWS Lambda function is working, it’s time to automate this and have it run every hour. We’ll pick different words because we know those exist. Let’s pretend we want to know when my personal website will be updated with the words “swift, java, and angular.” 50 | 51 | For this we’ll be using AWS CloudWatch events. So let’s head over there and create a new events rule. 52 | 53 | ![](https://cdn-images-1.medium.com/max/1600/1*7D-BzX42lIgLyQLwpywXRw.png) 54 | 55 | First we set the schedule to a fixed rate of 1 hour. Next, choose the Lambda function we created earlier. And finally, the most important part, select Configure input > Constant (JSON text), and paste in the JSON with the data to send to our function (see code below). 56 | 57 | 58 | 59 | Once you fill that in, click Configure details to name the rule and then create it. We now have the script running every hour, scraping our website, searching for the text we provided, and alerting us when it finds it. 60 | 61 | -------------------------------------------------------------------------------- /2018年6月/20180622 第262期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180622 第262期/README.md -------------------------------------------------------------------------------- /2018年6月/20180622 第263期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180622 第263期/README.md -------------------------------------------------------------------------------- /2018年6月/20180623 第264期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180623 第264期/README.md -------------------------------------------------------------------------------- /2018年6月/20180624 第265期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180624 第265期/README.md -------------------------------------------------------------------------------- /2018年6月/20180626 第266期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180626 第266期/README.md -------------------------------------------------------------------------------- /2018年6月/20180626 第267期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180626 第267期/README.md -------------------------------------------------------------------------------- /2018年6月/20180628 第268期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180628 第268期/README.md -------------------------------------------------------------------------------- /2018年6月/20180628 第269期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180628 第269期/README.md -------------------------------------------------------------------------------- /2018年6月/20180629 第270期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180629 第270期/README.md -------------------------------------------------------------------------------- /2018年6月/20180630 第271期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年6月/20180630 第271期/README.md -------------------------------------------------------------------------------- /2018年7月/20180701 第272期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180701 第272期/README.md -------------------------------------------------------------------------------- /2018年7月/20180702 第273期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180702 第273期/README.md -------------------------------------------------------------------------------- /2018年7月/20180703 第274期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180703 第274期/README.md -------------------------------------------------------------------------------- /2018年7月/20180704 第275期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180704 第275期/README.md -------------------------------------------------------------------------------- /2018年7月/20180705 第276期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180705 第276期/README.md -------------------------------------------------------------------------------- /2018年7月/20180707 第277期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180707 第277期/README.md -------------------------------------------------------------------------------- /2018年7月/20180707 第278期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180707 第278期/README.md -------------------------------------------------------------------------------- /2018年7月/20180708 第279期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180708 第279期/README.md -------------------------------------------------------------------------------- /2018年7月/20180709 第280期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180709 第280期/README.md -------------------------------------------------------------------------------- /2018年7月/20180710 第281期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180710 第281期/README.md -------------------------------------------------------------------------------- /2018年7月/20180711 第282期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180711 第282期/README.md -------------------------------------------------------------------------------- /2018年7月/20180713 第283期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180713 第283期/README.md -------------------------------------------------------------------------------- /2018年7月/20180713 第284期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180713 第284期/README.md -------------------------------------------------------------------------------- /2018年7月/20180714 第285期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180714 第285期/README.md -------------------------------------------------------------------------------- /2018年7月/20180716 第286期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180716 第286期/README.md -------------------------------------------------------------------------------- /2018年7月/20180716 第287期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180716 第287期/README.md -------------------------------------------------------------------------------- /2018年7月/20180717 第288期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180717 第288期/README.md -------------------------------------------------------------------------------- /2018年7月/20180718 第289期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180718 第289期/README.md -------------------------------------------------------------------------------- /2018年7月/20180719 第290期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180719 第290期/README.md -------------------------------------------------------------------------------- /2018年7月/20180721 第291期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180721 第291期/README.md -------------------------------------------------------------------------------- /2018年7月/20180722 第292期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180722 第292期/README.md -------------------------------------------------------------------------------- /2018年7月/20180722 第293期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180722 第293期/README.md -------------------------------------------------------------------------------- /2018年7月/20180723 第294期/README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/2018年7月/20180723 第294期/README.md -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # HackCV网站文章翻译 2 | 3 | 网站：https://hackcv.com/ 4 | 5 | > HackCV致力于收集、分享最新的机器学习、人工智能和编程技巧相关的内容。建立的初衷是收集真正有料的资讯和博客，整理优秀的代码框架和模型，每天分享自己的学习心得，与大家一起体验这波人工智能对社会生活带来的影响 6 | 7 | - [微博](https://weibo.com/hackcv) 8 | - [Twitter](https://twitter.com/hackcv) 9 | 10 | 翻译已取得站长本人许可，持续翻译ing。。。 11 | 12 | 翻译工具：我们骄傲地使用谷歌翻译、有道翻译、百度翻译等工具进行翻译辅助。 13 | 14 | 若翻译的文章中有侵犯到文章作者本人的权益，请联系我们。 15 | 16 | ## 翻译贡献者 17 | 18 | | 日期 | 翻译 | 校对 | 19 | | --------------------------------------------------- | ------------------------------------------------------------ | ---- | 20 | | [2017/09/25 第1期](https://hackcv.com/daily/p/1/) | [@wnma](https://github.com/wnma3mz) | | 21 | | [2017/10/04 第2期](https://hackcv.com/daily/p/2/) | [@doordiey](https://github.com/doordiey) | | 22 | | [2017/10/05 第3期](https://hackcv.com/daily/p/3/) | [@Arron206](https://github.com/Arron206) | | 23 | | [2017/10/06 第4期](https://hackcv.com/daily/p/4/) | [@mllove](https://github.com/mllove) | | 24 | | [2017/10/07 第5期](https://hackcv.com/daily/p/5/) | [@wnma](https://github.com/wnma3mz) | | 25 | | [2017/10/08 第6期](https://hackcv.com/daily/p/6/) | [@doordiey](https://github.com/doordiey) | | 26 | | [2017/10/09 第7期](https://hackcv.com/daily/p/7/) | [@mllove](https://github.com/mllove) | | 27 | | [2017/10/10 第8期](https://hackcv.com/daily/p/8/) | [@AlexdanerZe](https://github.com/AlexdanerZe) | | 28 | | [2017/10/11 第9期](https://hackcv.com/daily/p/9/) | [@exqlnet](https://github.com/exqlnet) | | 29 | | [2017/10/11 第10期](https://hackcv.com/daily/p/10/) | [@aboutmydreams](https://github.com/aboutmydreams) | | 30 | | [2017/10/13 第11期](https://hackcv.com/daily/p/11/) | [@wnma](https://github.com/wnma3mz) | | 31 | | [2017/10/14 第12期](https://hackcv.com/daily/p/12/) | [@wnma](https://github.com/wnma3mz) | | 32 | | [2017/10/15 第13期](https://hackcv.com/daily/p/13/) | | | 33 | | [2017/10/16 第14期](https://hackcv.com/daily/p/14/) | [@pickonecat](https://github.com/pickonecat)（暂未完成） | | 34 | | [2017/10/17 第15期](https://hackcv.com/daily/p/15/) | | | 35 | | [2017/10/18 第16期](https://hackcv.com/daily/p/16/) | | | 36 | | [2017/10/19 第17期](https://hackcv.com/daily/p/17/) | [@lbllol365](https://github.com/lbllol365?tdsourcetag=s_pctim_aiomsg) | | 37 | | [2017/10/20 第18期](https://hackcv.com/daily/p/18/) | | | 38 | 39 | ## 贡献指南 40 | 41 | 项目负责人：[wnma](https://github.com/wnma3mz)，QQ：1003324213 42 | 43 | 当然也欢迎直接提交pr。 44 | 45 | > 请您勇敢地去翻译和改进翻译。虽然我们追求卓越，但我们并不要求您做到十全十美，因此请不要担心因为翻译上犯错——在大部分情况下，我们的服务器已经记录所有的翻译，因此您不必担心会因为您的失误遭到无法挽回的破坏。（改编自维基百科） 46 | 47 | -------------------------------------------------------------------------------- /trans-docx/translated-.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-1.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-1.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-10 tips on using Jupyter Notebook.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-10 tips on using Jupyter Notebook.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-10-fearless-predictions-for-deep-learning-in-2018.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-10-fearless-predictions-for-deep-learning-in-2018.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-100-times-faster-natural-language-processing-in-python.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-100-times-faster-natural-language-processing-in-python.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-11行python实现神经网络.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-11行python实现神经网络.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-12-useful-things-to-know-about-machine-learning.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-12-useful-things-to-know-about-machine-learning.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-13-deep-learning-frameworks-for-natural-language-processing-in-python.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-13-deep-learning-frameworks-for-natural-language-processing-in-python.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-5-ways-of-looking-at-machine-learning-what-is-ml.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-5-ways-of-looking-at-machine-learning-what-is-ml.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-50-must-read-books-for-machine-learning.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-50-must-read-books-for-machine-learning.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-8 Essential Tips for People starting a Career in Data Science.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-8 Essential Tips for People starting a Career in Data Science.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-8-deep-learning-best-practices-i-learned-about-in-2017.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-8-deep-learning-best-practices-i-learned-about-in-2017.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Advice For New and Junior Data Scientists.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Advice For New and Junior Data Scientists.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-AlphaGo与能力增强.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-AlphaGo与能力增强.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Approaching (Almost) Any Machine Learning Problem Abhishek Thakur.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Approaching (Almost) Any Machine Learning Problem Abhishek Thakur.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Batch normalization in Neural Networks.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Batch normalization in Neural Networks.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Bayesian Learning for Statistical Classification.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Bayesian Learning for Statistical Classification.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Best Practices for Document Classification with Deep Learning.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Best Practices for Document Classification with Deep Learning.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Fundamentals of Deep Learning – Activation Functions and When to Use Them.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Fundamentals of Deep Learning – Activation Functions and When to Use Them.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Google Says Machine Learning Chips Make AI Faster and More Efficient.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Google Says Machine Learning Chips Make AI Faster and More Efficient.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Hey Siri An On-device DNN-powered Voice Trigger for Apple’s Personal Assistant.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Hey Siri An On-device DNN-powered Voice Trigger for Apple’s Personal Assistant.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-How to Prepare Movie Review Data for Sentiment Analysis.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-How to Prepare Movie Review Data for Sentiment Analysis.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Instacart Market Basket Analysis, Winner's Interview 2nd place, Kazuki Onodera.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Instacart Market Basket Analysis, Winner's Interview 2nd place, Kazuki Onodera.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Intel® Nervana™ Neural Network Processors (NNP) Redefine AI Silicon.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Intel® Nervana™ Neural Network Processors (NNP) Redefine AI Silicon.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Natural Stupidity is more Dangerous than Artificial Intelligence.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Natural Stupidity is more Dangerous than Artificial Intelligence.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Planet Understanding the Amazon from Space, 1st Place Winner's Interview.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Planet Understanding the Amazon from Space, 1st Place Winner's Interview.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Speech Recognition Is Not Solved.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Speech Recognition Is Not Solved.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Subword-level embeddings.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Subword-level embeddings.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-The difference between Statistical Modeling and Machine Learning, as I see it.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-The difference between Statistical Modeling and Machine Learning, as I see it.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Why AlphaGo Zero is a Quantum Leap Forward in Deep Learning.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Why AlphaGo Zero is a Quantum Leap Forward in Deep Learning.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Word Tensors.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Word Tensors.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-Word embeddings in 2017 Trends and future directions.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-Word embeddings in 2017 Trends and future directions.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-a-deeper-understanding-of-nnets-part-3-lstm-and-gru.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-a-deeper-understanding-of-nnets-part-3-lstm-and-gru.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-a-genetic-algorithm-scheduled-our-developer-conference-heres-what-i-learnt.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-a-genetic-algorithm-scheduled-our-developer-conference-heres-what-i-learnt.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-a-simple-spell-checker-built-from-word-vectors.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-a-simple-spell-checker-built-from-word-vectors.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-adding-a-cutting-edge-deep-learning-training-technique-to-the-fast-ai-library.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-adding-a-cutting-edge-deep-learning-training-technique-to-the-fast-ai-library.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-an-introduction-to-biomedical-image-analysis-with-tensorflow-and-dltk.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-an-introduction-to-biomedical-image-analysis-with-tensorflow-and-dltk.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-apple-ai-interview-questions-acing-the-ai-interview.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-apple-ai-interview-questions-acing-the-ai-interview.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-artificial-intelligence-the-revolution-hasnt-happened-yet.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-artificial-intelligence-the-revolution-hasnt-happened-yet.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-artwork-personalization.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-artwork-personalization.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-asking-the-right-questions-about-ai.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-asking-the-right-questions-about-ai.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-ava-the-art-and-science-of-image-discovery-at-netflix.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-ava-the-art-and-science-of-image-discovery-at-netflix.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-benchmarking-tensorflow-performance-on-next-generation-gpus.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-benchmarking-tensorflow-performance-on-next-generation-gpus.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-can-deep-networks-learn-invariants.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-can-deep-networks-learn-invariants.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-colab-an-easy-way-to-learn-and-use-tensorflow.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-colab-an-easy-way-to-learn-and-use-tensorflow.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-collecting-weather-data-to-boost-data-science-models-with-selenium.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-collecting-weather-data-to-boost-data-science-models-with-selenium.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-comprehensive-guide-to-generative-adversarial-networks-and-wasserstein-gans.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-comprehensive-guide-to-generative-adversarial-networks-and-wasserstein-gans.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-cvpr-2018-impressions-and-15-interesting-papers.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-cvpr-2018-impressions-and-15-interesting-papers.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-dcnet-denoising-dna-sequence-with-a-lstm-rnn-and-pytorch.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-dcnet-denoising-dna-sequence-with-a-lstm-rnn-and-pytorch.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-deep-learning-and-time-to-predict-emojis.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-deep-learning-and-time-to-predict-emojis.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-ethics-in-machine-learning.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-ethics-in-machine-learning.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-exploring-deep-learning-and-neural-network-modeler-with-watson-studio.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-exploring-deep-learning-and-neural-network-modeler-with-watson-studio.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-first-time-with-kaggle-a-convnet-to-classify-toxic-comments-with-keras.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-first-time-with-kaggle-a-convnet-to-classify-toxic-comments-with-keras.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-from-research-to-practice.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-from-research-to-practice.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-from-zero-to-research-an-introduction-to-meta-learning.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-from-zero-to-research-an-introduction-to-meta-learning.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-getting-the-most-of-xgboost-and-lightgbm-speed-compiler-cpu-pinning.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-getting-the-most-of-xgboost-and-lightgbm-speed-compiler-cpu-pinning.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-growing-up-with-ai-how-can-families-play-and-learn-with-their-new-smart-toys-and-companions.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-growing-up-with-ai-how-can-families-play-and-learn-with-their-new-smart-toys-and-companions.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-highlights-from-tensorflow-developer-summit-2018.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-highlights-from-tensorflow-developer-summit-2018.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-how-to-build-a-neural-network-framework-like-tensorflow-in-c-part-1.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-how-to-build-a-neural-network-framework-like-tensorflow-in-c-part-1.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-how-to-build-your-own-alphazero-ai-using-python-and-keras.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-how-to-build-your-own-alphazero-ai-using-python-and-keras.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-how-to-build-your-own-world-model-using-python-and-keras.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-how-to-build-your-own-world-model-using-python-and-keras.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-how-to-create-a-machine-learning-framework-from-scratch-in-491-steps.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-how-to-create-a-machine-learning-framework-from-scratch-in-491-steps.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-how-to-easily-detect-objects-with-deep-learning-on-raspberrypi.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-how-to-easily-detect-objects-with-deep-learning-on-raspberrypi.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-how-to-use-deep-learning-when-you-have-limited-data-part-2-data-augmentation.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-how-to-use-deep-learning-when-you-have-limited-data-part-2-data-augmentation.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-how-we-grew-from-0-to-4-million-women-on-our-fashion-app-with-a-vertical-machine-learning-approach.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-how-we-grew-from-0-to-4-million-women-on-our-fashion-app-with-a-vertical-machine-learning-approach.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-https-medium-com-100-shades-of-machine-learning-rediscovering-semantic-segmentation-part1.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-https-medium-com-100-shades-of-machine-learning-rediscovering-semantic-segmentation-part1.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-icml-2018-notes.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-icml-2018-notes.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-install-tensorflow-1-8-0-with-gpu-from-source-on-ubuntu-18-04-bionic-beaver.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-install-tensorflow-1-8-0-with-gpu-from-source-on-ubuntu-18-04-bionic-beaver.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-instance-embedding-instance-segmentation-without-proposals.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-instance-embedding-instance-segmentation-without-proposals.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-introducing-prose-v2-0-0-bringing-nlp-to-go.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-introducing-prose-v2-0-0-bringing-nlp-to-go.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-introducing-tensorflow-hub-a-library-for-reusable-machine-learning-modules-in-tensorflow.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-introducing-tensorflow-hub-a-library-for-reusable-machine-learning-modules-in-tensorflow.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-introducing-tensorflow-js-machine-learning-in-javascript.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-introducing-tensorflow-js-machine-learning-in-javascript.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-its-a-no-brainer-deep-learning-for-brain-mr-images.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-its-a-no-brainer-deep-learning-for-brain-mr-images.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-logistic-regression-vs-deep-neural-networks-david.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-logistic-regression-vs-deep-neural-networks-david.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-maximum-likelihood-maximum-a-priori-and-bayesian-parameter-estimation.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-maximum-likelihood-maximum-a-priori-and-bayesian-parameter-estimation.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-meltdown.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-meltdown.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-mltest-automatically-test-neural-network-models-in-one-function-call.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-mltest-automatically-test-neural-network-models-in-one-function-call.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-move-mirror-an-ai-experiment-with-pose-estimation-in-the-browser-using-tensorflow-js.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-move-mirror-an-ai-experiment-with-pose-estimation-in-the-browser-using-tensorflow-js.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-multi-gpu-rosetta-stone.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-multi-gpu-rosetta-stone.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-multi-modal-methods-part-one.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-multi-modal-methods-part-one.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-mxboard-mxnet-data-visualization.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-mxboard-mxnet-data-visualization.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-mxnet-for-pytorch-users-in-10-minutes.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-mxnet-for-pytorch-users-in-10-minutes.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-ncrf-deep-learning-framework-improves-cancer-metastasis-detection.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-ncrf-deep-learning-framework-improves-cancer-metastasis-detection.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-neural-network-learning-part-1-feed-forward.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-neural-network-learning-part-1-feed-forward.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-neuronuggets-age-and-gender-estimation.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-neuronuggets-age-and-gender-estimation.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-nips-2017-policy-field-notes.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-nips-2017-policy-field-notes.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-only-numpy-why-i-do-manual-back-propagation-implementing-multi-channel-layer-convolution-neural.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-only-numpy-why-i-do-manual-back-propagation-implementing-multi-channel-layer-convolution-neural.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-open-machine-learning-course-topic-10-gradient-boosting.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-open-machine-learning-course-topic-10-gradient-boosting.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-open-machine-learning-course-topic-3-classification-decision-trees-and-k-nearest-neighbors.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-open-machine-learning-course-topic-3-classification-decision-trees-and-k-nearest-neighbors.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-open-machine-learning-course-topic-9-time-series-analysis-in-python.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-open-machine-learning-course-topic-9-time-series-analysis-in-python.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-otodscinos-the-root-cause-of-slow-neural-net-training.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-otodscinos-the-root-cause-of-slow-neural-net-training.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-predicting-the-price-of-wine-with-the-keras-functional-api-and-tensorflow.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-predicting-the-price-of-wine-with-the-keras-functional-api-and-tensorflow.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-real-time-human-pose-estimation-in-the-browser-with-tensorflow-js.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-real-time-human-pose-estimation-in-the-browser-with-tensorflow-js.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-self-driving-cars-are-here.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-self-driving-cars-are-here.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-sentiment-analysis-torchtext.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-sentiment-analysis-torchtext.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-sentiment-classification-from-keras-to-the-browser.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-sentiment-classification-from-keras-to-the-browser.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-simple-website-text-scraping-with-go-and-aws-lambda.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-simple-website-text-scraping-with-go-and-aws-lambda.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-snapchats-filters-how-computer-vision-recognizes-your-face.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-snapchats-filters-how-computer-vision-recognizes-your-face.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-speed-up-tensorflow-inference-on-gpus-with-tensorrt.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-speed-up-tensorflow-inference-on-gpus-with-tensorrt.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-super-simple-end-to-end-test-of-keras-tensorflow-and-coreml.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-super-simple-end-to-end-test-of-keras-tensorflow-and-coreml.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-tensorflow-cpus-and-gpus-configuration.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-tensorflow-cpus-and-gpus-configuration.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-tensorflow-estimator-dataset-apis.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-tensorflow-estimator-dataset-apis.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-the-3-tricks-that-made-alphago-zero-work.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-the-3-tricks-that-made-alphago-zero-work.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-the-8-neural-network-architectures-machine-learning-researchers-need-to-learn.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-the-8-neural-network-architectures-machine-learning-researchers-need-to-learn.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-the-age-of-personal-assistants-more-machine-learning-less-hand-crafting.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-the-age-of-personal-assistants-more-machine-learning-less-hand-crafting.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-the-end-of-video-coding.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-the-end-of-video-coding.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-the-two-steps-people-forget-to-take-when-applying-machine-learning.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-the-two-steps-people-forget-to-take-when-applying-machine-learning.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-things-we-wish-we-had-known-before-we-started-our-first-machine-learning-project.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-things-we-wish-we-had-known-before-we-started-our-first-machine-learning-project.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-toward-a-practical-perceptual-video-quality-metric.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-toward-a-practical-perceptual-video-quality-metric.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-transfer-learning-vs-multitask-ibrahim.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-transfer-learning-vs-multitask-ibrahim.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-transfer-learning-will-radically-change-machine-learning-for-engineers.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-transfer-learning-will-radically-change-machine-learning-for-engineers.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-universal-word-sentence-embeddings.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-universal-word-sentence-embeddings.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-using-deep-learning-to-automatically-rank-millions-of-hotel-images.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-using-deep-learning-to-automatically-rank-millions-of-hotel-images.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-what-do-we-learn-from-region-based-object-detectors-faster-r-cnn-r-fcn-fpn.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-what-do-we-learn-from-region-based-object-detectors-faster-r-cnn-r-fcn-fpn.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-what-was-ai-in-2017.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-what-was-ai-in-2017.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-why-you-should-start-using-in-place-activated-batchnorm.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-why-you-should-start-using-in-place-activated-batchnorm.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-使用YOLO提高实时目标检测能力.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-使用YOLO提高实时目标检测能力.md.docx -------------------------------------------------------------------------------- /trans-docx/translated-如何使用C++写出一个TensorFlow.md.docx: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apachecn/HackCV-Translate/6987895a7c8d47f72b974dd31fd4f670b2f3d5ca/trans-docx/translated-如何使用C++写出一个TensorFlow.md.docx --------------------------------------------------------------------------------