├── dl ├── dl.md ├── rbm │ ├── dbn.md │ └── rbm-yuanli.md ├── introduction │ ├── READ.md │ └── chapter1.md ├── rnn │ ├── deep-bidirectional-rnns.md │ ├── bidirectional-rnns.md │ └── rnn.md ├── encoder │ ├── stack-denoise-encoder.md │ └── sparse-autoencoder.md ├── reinforcement │ ├── policy-network.md │ ├── a3csuan-fa.md │ ├── ppo.md │ ├── banditsuan-fa.md │ ├── actor-critic.md │ ├── ddpg.md │ ├── policy-gridient.md │ ├── dqn.md │ ├── q-learning.md │ ├── multi-steps-bootstraping.md │ ├── ti-du-xia-jiang-ce-lve-you-hua-tui-dao.md │ ├── monte-carlo.md │ └── alpha-beta.md ├── popnet │ ├── resnet.md │ └── mobilenet.md ├── layers │ ├── wrapper.md │ ├── layers.md │ ├── lcnn.md │ ├── merge.md │ ├── nosie.md │ ├── core.md │ ├── userdefine.md │ ├── rnn.md │ ├── regular.md │ ├── ebbedded.md │ └── conv.md └── word2vec │ └── word2vec.md ├── ml ├── ml.md ├── lda.md ├── regression │ ├── kl.md │ └── regression.md ├── clean-feature │ ├── README.md │ ├── outlier-detect.md │ ├── one-hot.md │ └── scikit-lda.md ├── evolution │ ├── evolution.md │ ├── evolution-strategy.md │ ├── neat.md │ └── yichuansuanfa.md ├── associative │ └── associative.md ├── lda │ └── lda.md ├── bayes │ ├── bian-fen-bei-xie-si-tui-duan.md │ └── simple-bayes-real-use.md ├── pythonml │ ├── ji-qi-xue-xi-ku-ff08-tensorflow-keras.md │ ├── pythonji-qi-shu-xue-ku.md │ ├── mo-xing-xuan-ze.md │ └── ml-metrics.md ├── cluster.md ├── hmm │ └── markv-mengtekluo.md ├── decisiontree.md ├── crf │ └── crf.md ├── cluster │ ├── gmm.md │ ├── knn-code.md │ ├── ap.md │ └── knnshi-jian.md ├── recommand │ ├── mr-itemcf.md │ └── fm.md └── forest.md ├── can-kao.md ├── nlp ├── nlp.md └── lda │ └── lsi.md ├── jin-hua-suan-fa.md ├── images ├── gan │ ├── README.md │ ├── vae1.png │ ├── vae2.png │ ├── vae3.png │ ├── vae4.png │ ├── vae5.png │ ├── vae6.png │ ├── vae7.png │ ├── vae8.png │ ├── vae9.png │ ├── vae10.png │ └── vae11.png ├── 1.png ├── 2.png ├── 3.png ├── 4.png ├── 5.png ├── 6.png ├── 7.png ├── 8.png ├── 9.png ├── 10.png ├── 100.png ├── 101.png ├── 102.png ├── 103.png ├── 104.png ├── 105.png ├── 106.png ├── 107.png ├── 108.png ├── 109.png ├── 11.png ├── 110.png ├── 111.png ├── 112.png ├── 113.png ├── 114.png ├── 115.png ├── 116.png ├── 117.png ├── 118.png ├── 119.png ├── 12.png ├── 120.png ├── 121.png ├── 122.png ├── 123.png ├── 124.png ├── 125.png ├── 126.png ├── 127.png ├── 128.png ├── 129.png ├── 13.png ├── 130.png ├── 131.png ├── 132.png ├── 133.png ├── 134.png ├── 135.png ├── 136.png ├── 137.png ├── 138.png ├── 139.png ├── 14.png ├── 140.png ├── 141.png ├── 142.png ├── 143.png ├── 144.png ├── 145.png ├── 146.png ├── 147.png ├── 148.png ├── 149.png ├── 15.png ├── 150.png ├── 151.png ├── 152.png ├── 153.png ├── 154.png ├── 155.png ├── 156.png ├── 157.png ├── 158.png ├── 159.png ├── 16.png ├── 160.png ├── 161.png ├── 162.png ├── 163.png ├── 164.png ├── 165.png ├── 17.png ├── 18.png ├── 19.png ├── 20.png ├── 21.png ├── 22.png ├── 23.png ├── 24.png ├── 25.png ├── 26.png ├── 27.png ├── 28.png ├── 29.png ├── 30.png ├── 31.png ├── 32.png ├── 33.png ├── 34.png ├── 35.png ├── 36.png ├── 37.png ├── 38.png ├── 39.png ├── 40.png ├── 41.png ├── 42.png ├── 43.png ├── 44.png ├── 45.png ├── 46.png ├── 47.png ├── 48.png ├── 49.png ├── 50.png ├── 51.png ├── 52.png ├── 53.png ├── 54.png ├── 55.png ├── 56.png ├── 57.png ├── 58.png ├── 59.png ├── 60.png ├── 61.png ├── 62.png ├── 63.png ├── 64.png ├── 65.png ├── 66.png ├── 67.png ├── 68.png ├── 69.png ├── 70.png ├── 71.png ├── 72.png ├── 73.png ├── 74.png ├── 75.png ├── 76.png ├── 77.png ├── 78.png ├── 79.png ├── 80.png ├── 81.png ├── 82.png ├── 83.png ├── 84.png ├── 85.png ├── 86.png ├── 87.png ├── 88.png ├── 89.png ├── 90.png ├── 91.png ├── 92.png ├── 93.png ├── 94.png ├── 95.png ├── 96.png ├── 97.png ├── 98.png ├── 99.png ├── svm │ └── 1.png ├── bp │ ├── 0111.png │ ├── 01114321.png │ ├── qwerqwreqerqer.png │ ├── qewrqwreqrwerqwer.png │ ├── weqrqwrcqwerqwer.png │ ├── qwerqrqwreqwerqwere.png │ └── weqrqwrcqqwerqwrcwerqwer.png ├── overview.png ├── dl │ └── gan │ │ ├── clip1.png │ │ ├── clip2.png │ │ ├── import.png │ │ ├── minmax.png │ │ ├── table1.png │ │ ├── gan3loss.png │ │ └── import2.png ├── cluster │ ├── table1.png │ ├── table2.png │ ├── table3.png │ └── table4.png ├── cnn │ └── weqrqwrcqwerqwer.png ├── machine-learning-classification.jpg ├── overfitting │ ├── 20151102210633783.png │ ├── 20151102210852970.png │ ├── 20151102211204226.png │ ├── 20151102211214508.png │ └── 20151102211315525.png └── metrics │ ├── 1042406-20161024154443875-2037260202.jpg │ └── 1042406-20161024164359046-1869944207.png ├── resources ├── introduction.md └── dl │ └── list.md ├── datasets ├── readme.md ├── Data.csv ├── studentscores.csv └── 50_Startups.csv ├── math ├── linear-matrix │ ├── tezhengzhihetezhengxiangliang.md │ └── lapack.md ├── analytic │ ├── lagelangri-kkt.md │ ├── tuyouhua.md │ ├── cross-validation.md │ └── common-function.md ├── probability │ └── mle.md └── math.md ├── cover.jpg ├── GLOSSARY.md ├── assets ├── dlnet2.png ├── dlnet3.png ├── dlnet4.png ├── dlnet5.png ├── dlnet6.png ├── gif5.gif ├── ica2.png ├── ica3.png ├── mlpre1.png ├── mlpre2.png ├── mlpre3.png ├── mlpre4.png ├── mlpre5.png ├── mlpre6.png ├── mlpre7.png ├── mlpre8.png ├── mlpre9.png ├── rl-td1.png ├── rl-td2.png ├── rl-td4.png ├── rl-td48png ├── rl-td5.png ├── rnnb1.png ├── rnnb2.png ├── rnnb3.png ├── ubc1.png ├── ucts.png ├── vi-em.png ├── 5_13_03.gif ├── 5_13_06.gif ├── cnnconv1.png ├── cnnconv2.png ├── cnnconv3.png ├── cnnpool1.png ├── dlnet22.png ├── dlnet23.png ├── dlnet24.png ├── dnnimg1.png ├── dnnimg2.png ├── dnnimg3.png ├── dnnimg4.png ├── dnnimg5.png ├── dnnimg6.png ├── dnnimg7.png ├── encode4.png ├── encoder1.png ├── encoder2.png ├── encoder3.png ├── encoder5.png ├── encoder7.png ├── ganvae1.png ├── ganvae2.png ├── ganvae3.png ├── ganvae4.png ├── ganvae5.png ├── lg-kkt1.png ├── lg-kkt2.png ├── lg-kkt3.png ├── lg-kkt4.png ├── lg-kkt5.png ├── lstmpci4.png ├── lstmpic1.png ├── lstmpic2.png ├── lstmpic3.png ├── lstmpic5.png ├── lstmpic6.png ├── lstmpic7.png ├── lstmpic8.png ├── mdp-dp1.png ├── mdp-dp2.png ├── meituan1.png ├── meituan2.png ├── meituan3.png ├── meituan4.png ├── meituan5.png ├── meituan7.png ├── meituan8.png ├── meituan9.png ├── mlpre11.png ├── mlpre12.png ├── mlpre13.png ├── mlpre14.png ├── mlpre15.png ├── mlpre16.png ├── mlpre17.png ├── newton1.png ├── restnet1.png ├── rl-td11.png ├── rl-td21.png ├── rl-td31.png ├── rl-td33.png ├── rl-td34.png ├── rl-td41.png ├── rl-td46.png ├── rl-td47.png ├── rl-td48.png ├── rl-td49.png ├── rl-td51.png ├── rl-td53.png ├── rl-td61.png ├── rl-td71.png ├── rl-td81.png ├── softmax1.png ├── softmax2.png ├── softmax3.png ├── cnnalexnet1.png ├── cnndensnet1.png ├── cnndensnet2.png ├── convetex1.png ├── convetex2.png ├── convetex3.png ├── dlintronet2.png ├── gif5新文件 (1).gif ├── googlenet1.png ├── hmm-markv3.png ├── hmm-markv4.png ├── hmm-markv5.png ├── intronet1.png ├── jiekadejuli.png ├── knncode100.jpg ├── knncode100.png ├── markv-mdp1.png ├── markv-mdp11.png ├── markv-mdp2.png ├── markv-mrp1.png ├── meituan10.png ├── meituan11.png ├── meituan12.png ├── meituan14.png ├── meituan15.png ├── meituan16.png ├── multi-bt1.png ├── multi-bt10.png ├── multi-bt11.png ├── multi-bt12.png ├── multi-bt2.png ├── multi-bt3.png ├── multi-bt5.png ├── multi-bt7.png ├── multi-bt8.png ├── multi-bt9.png ├── mywebchat.png ├── prefixspan1.png ├── prefixspan2.png ├── prefixspan3.png ├── prefixspan4.png ├── prefixspan5.png ├── rbm-yuanli2.png ├── rbm-yuanli3.png ├── rbm-yuanli5.png ├── rbm-yuanli6.png ├── rbm-yuanli7.png ├── rbm-yuanli8.png ├── rbm-yuanli9.png ├── rbm-yunali1.png ├── rbm-yunali4.png ├── rl-td-error.png ├── runseqtask1.png ├── simplenet.png ├── touyouhua10.png ├── touyouhua11.png ├── touyouhua2.png ├── touyouhua3.png ├── touyouhua4.png ├── touyouhua5.png ├── touyouhua6.png ├── touyouhua7.png ├── touyouhua8.png ├── touyouhua9.png ├── touyuhua1.png ├── xinxishang.png ├── alirecomadn7.png ├── alirecomamnd6.png ├── alirecommadn4.png ├── alirecommand1.png ├── alirecommand2.png ├── alirecommand5.png ├── alirecoomand3.png ├── commonf-rleu.png ├── commonf-tanh.png ├── deeplayer-bn1.png ├── deeplayer-bn2.png ├── deeplayer-bn3.png ├── makrv-bestvf.png ├── markv-mdpvf1.png ├── markv-policy1.png ├── markv-reward1.png ├── mento-calro2.png ├── mento-carlo1.png ├── mento-carlo3.png ├── mento-carlo4.png ├── mento-carlo5.png ├── mento-carlo6.png ├── mento-carlo7.png ├── mento-carlo8.png ├── rbm-yuanli10.png ├── rbm-yuanli11.png ├── rbm-yuanli21.png ├── rbm-yuanli22.png ├── rbm-yuanli23.png ├── rbm-yuanli24.png ├── rbm-yuanli25.png ├── rnnstructure1.png ├── voice-intro1.png ├── voice-intro2.png ├── voice-intro3.png ├── voice-intro4.png ├── voice-intro5.png ├── voice-intro6.png ├── voice-intro7.png ├── clustermethod1.png ├── clustermethod2.png ├── deeplearinggan4.png ├── deeplearinggan5.png ├── deeplearinggan6.png ├── deeplearinggan8.png ├── deeplearninggan1.png ├── deeplearninggan2.png ├── deeplearninggan3.png ├── deeplearninggan7.png ├── deeplearninggan9.png ├── dl-cnn-for-nlp.png ├── hmm-mengtekaluo2.png ├── hmm-mentekaluo1.png ├── label-recommand.png ├── layers-softmax.png ├── logisticcode100.jpg ├── markv-bellman1.png ├── markv-bellman2.png ├── markv-bellman3.png ├── mlmodelmetrics10.png ├── mlmodelmetrics11.png ├── mlmodelmetrics12.png ├── mlmodelmetrics13.png ├── mlmodelmetrics14.png ├── mlmodelmetrics15.png ├── mlmodelmetrics16.png ├── mlmodelmetrics17.png ├── mlmodelmetrics2.png ├── mlmodelmetrics3.png ├── mlmodelmetrics32.png ├── mlmodelmetrics6.png ├── mlmodelmetrics8.png ├── mlmodelmetrics9.png ├── mlmodemetrics1.png ├── mlmodemetrics18.png ├── mlmodemetrics19.png ├── mlmodemetrics7.png ├── multibandit-gba.png ├── multibandit-gba2.png ├── multilrcode100.jpg ├── mutilbandit-ucb.png ├── reinforcement-ql.png ├── reinforcemnt-ql4.png ├── runnbidirection1.png ├── simplelrcode100.jpg ├── dataprocesscode100.jpg ├── deeplayer-pooling1.png ├── deeplearninggan10.png ├── deeplearninggan11.png ├── deeplearninggan14.png ├── deeplearninggan17.png ├── dl-popnet-resnet1.png ├── dl-popnet-resnet2.png ├── dl-popnet-resnet3.png ├── mlmodelmetircsd24.png ├── mlmodelmetricsd22.png ├── mlmodelmetricvs21.png ├── reinforcement-a3c.png ├── reinforcement-dqn1.png ├── reinforcement-dqn3.png ├── reinforcement-dqn4.png ├── reinforcement-dqn5.png ├── reinforcement-pg1.png ├── reinforcement-ql2.png ├── reinforcement-ql3.png ├── reinforcemnt-dqn2.png ├── decisiontreecode100.jpg ├── deeplayer-embeding2.png ├── deeplayer-embeding3.png ├── deeplayerr-embemding1.png ├── deeplearinginggan13.png ├── dl-poopnet-mobilenet1.png ├── dl-popnet-mobilenet.png ├── dl-popnet-mobilenet10.png ├── dl-popnet-mobilenet2.png ├── dl-popnet-mobilenet3.png ├── dl-popnet-mobilenet5.png ├── dl-popnet-mobilenet6.png ├── dl-popnet-mobilenet7.png ├── dl-popnet-mobilenet8.png ├── dl-popnet-mobilenet9.png ├── integrate-learning1.png ├── machinelearingdist2.png ├── machinelearningdist1.png ├── markv-transfer-matrix.png ├── markv-valuefunction1.png ├── markv-valuefunction2.png ├── markv-vaulefunction3.png ├── multi-bootstraping2.png ├── multibandit-formula1.png ├── multibandit-svgreward.png ├── normal-expandvaiance.png ├── randomforestcode100.jpg ├── reinforcement-dqn21.png ├── reinforcementlearing3.png ├── reinforcementlearing5.png ├── reinforementlearing7.png ├── reinforcementlearning1.png ├── reinforcementlearning2.png ├── reinforcementlearning4.png ├── reinforcementlearning6.png ├── reinforcemnt-opeai-ppo.png ├── mlutibandit-rewardreduce.png ├── multi-steps-bootstraping1.png ├── reinforcement-deepmind-ppo.png └── deterministic-value-iteration.png ├── styles ├── ebook.css ├── pdf.css └── website.css ├── .gitattributes ├── liu-xing-wang-luo-jie-gou.md ├── book.json ├── .gitignore ├── CONTRIBUTING.md └── mian-shi-ti-mu-1.md /dl/dl.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ml/ml.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /can-kao.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /dl/rbm/dbn.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ml/lda.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /nlp/nlp.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /jin-hua-suan-fa.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /dl/introduction/READ.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /images/gan/README.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ml/regression/kl.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ml/clean-feature/README.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ml/evolution/evolution.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /resources/introduction.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /dl/rnn/deep-bidirectional-rnns.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ml/associative/associative.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /dl/encoder/stack-denoise-encoder.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /dl/reinforcement/policy-network.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ml/evolution/evolution-strategy.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /ml/evolution/neat.md: -------------------------------------------------------------------------------- 1 | Neat算法介绍 2 | 3 | -------------------------------------------------------------------------------- /ml/lda/lda.md: -------------------------------------------------------------------------------- 1 | # 2 | 3 | 4 | 5 | -------------------------------------------------------------------------------- /ml/bayes/bian-fen-bei-xie-si-tui-duan.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /datasets/readme.md: -------------------------------------------------------------------------------- 1 | Day wise Dataset Used in Code 2 | -------------------------------------------------------------------------------- /ml/pythonml/ji-qi-xue-xi-ku-ff08-tensorflow-keras.md: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /math/linear-matrix/tezhengzhihetezhengxiangliang.md: -------------------------------------------------------------------------------- 1 | 2 | 3 | -------------------------------------------------------------------------------- /cover.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/cover.jpg -------------------------------------------------------------------------------- /GLOSSARY.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/GLOSSARY.md -------------------------------------------------------------------------------- /images/1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/1.png -------------------------------------------------------------------------------- /images/2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/2.png -------------------------------------------------------------------------------- /images/3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/3.png -------------------------------------------------------------------------------- /images/4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/4.png -------------------------------------------------------------------------------- /images/5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/5.png -------------------------------------------------------------------------------- /images/6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/6.png -------------------------------------------------------------------------------- /images/7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/7.png -------------------------------------------------------------------------------- /images/8.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/8.png -------------------------------------------------------------------------------- /images/9.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/9.png -------------------------------------------------------------------------------- /images/10.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/10.png -------------------------------------------------------------------------------- /images/100.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/100.png -------------------------------------------------------------------------------- /images/101.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/101.png -------------------------------------------------------------------------------- /images/102.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/102.png -------------------------------------------------------------------------------- /images/103.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/103.png -------------------------------------------------------------------------------- /images/104.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/104.png -------------------------------------------------------------------------------- /images/105.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/105.png -------------------------------------------------------------------------------- /images/106.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/106.png -------------------------------------------------------------------------------- /images/107.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/107.png -------------------------------------------------------------------------------- /images/108.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/108.png -------------------------------------------------------------------------------- /images/109.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/109.png -------------------------------------------------------------------------------- /images/11.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/11.png -------------------------------------------------------------------------------- /images/110.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/110.png -------------------------------------------------------------------------------- /images/111.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/111.png -------------------------------------------------------------------------------- /images/112.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/112.png -------------------------------------------------------------------------------- /images/113.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/113.png -------------------------------------------------------------------------------- /images/114.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/114.png -------------------------------------------------------------------------------- /images/115.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/115.png -------------------------------------------------------------------------------- /images/116.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/116.png -------------------------------------------------------------------------------- /images/117.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/117.png -------------------------------------------------------------------------------- /images/118.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/118.png -------------------------------------------------------------------------------- /images/119.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/119.png -------------------------------------------------------------------------------- /images/12.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/12.png -------------------------------------------------------------------------------- /images/120.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/120.png -------------------------------------------------------------------------------- /images/121.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/121.png -------------------------------------------------------------------------------- /images/122.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/122.png -------------------------------------------------------------------------------- /images/123.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/123.png -------------------------------------------------------------------------------- /images/124.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/124.png -------------------------------------------------------------------------------- /images/125.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/125.png -------------------------------------------------------------------------------- /images/126.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/126.png -------------------------------------------------------------------------------- /images/127.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/127.png -------------------------------------------------------------------------------- /images/128.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/128.png -------------------------------------------------------------------------------- /images/129.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/129.png -------------------------------------------------------------------------------- /images/13.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/13.png -------------------------------------------------------------------------------- /images/130.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/130.png -------------------------------------------------------------------------------- /images/131.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/131.png -------------------------------------------------------------------------------- /images/132.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/132.png -------------------------------------------------------------------------------- /images/133.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/133.png -------------------------------------------------------------------------------- /images/134.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/134.png -------------------------------------------------------------------------------- /images/135.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/135.png -------------------------------------------------------------------------------- /images/136.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/136.png -------------------------------------------------------------------------------- /images/137.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/137.png -------------------------------------------------------------------------------- /images/138.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/138.png -------------------------------------------------------------------------------- /images/139.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/139.png -------------------------------------------------------------------------------- /images/14.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/14.png -------------------------------------------------------------------------------- /images/140.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/140.png -------------------------------------------------------------------------------- /images/141.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/141.png -------------------------------------------------------------------------------- /images/142.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/142.png -------------------------------------------------------------------------------- /images/143.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/143.png -------------------------------------------------------------------------------- /images/144.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/144.png -------------------------------------------------------------------------------- /images/145.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/145.png -------------------------------------------------------------------------------- /images/146.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/146.png -------------------------------------------------------------------------------- /images/147.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/147.png -------------------------------------------------------------------------------- /images/148.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/148.png -------------------------------------------------------------------------------- /images/149.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/149.png -------------------------------------------------------------------------------- /images/15.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/15.png -------------------------------------------------------------------------------- /images/150.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/150.png -------------------------------------------------------------------------------- /images/151.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/151.png -------------------------------------------------------------------------------- /images/152.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/152.png -------------------------------------------------------------------------------- /images/153.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/153.png -------------------------------------------------------------------------------- /images/154.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/154.png -------------------------------------------------------------------------------- /images/155.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/155.png -------------------------------------------------------------------------------- /images/156.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/156.png -------------------------------------------------------------------------------- /images/157.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/157.png -------------------------------------------------------------------------------- /images/158.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/158.png -------------------------------------------------------------------------------- /images/159.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/159.png -------------------------------------------------------------------------------- /images/16.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/16.png -------------------------------------------------------------------------------- /images/160.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/160.png -------------------------------------------------------------------------------- /images/161.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/161.png -------------------------------------------------------------------------------- /images/162.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/162.png -------------------------------------------------------------------------------- /images/163.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/163.png -------------------------------------------------------------------------------- /images/164.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/164.png -------------------------------------------------------------------------------- /images/165.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/165.png -------------------------------------------------------------------------------- /images/17.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/17.png -------------------------------------------------------------------------------- /images/18.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/18.png -------------------------------------------------------------------------------- /images/19.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/19.png -------------------------------------------------------------------------------- /images/20.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/20.png -------------------------------------------------------------------------------- /images/21.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/21.png -------------------------------------------------------------------------------- /images/22.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/22.png -------------------------------------------------------------------------------- /images/23.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/23.png -------------------------------------------------------------------------------- /images/24.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/24.png -------------------------------------------------------------------------------- /images/25.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/25.png -------------------------------------------------------------------------------- /images/26.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/26.png -------------------------------------------------------------------------------- /images/27.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/27.png -------------------------------------------------------------------------------- /images/28.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/28.png -------------------------------------------------------------------------------- /images/29.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/29.png -------------------------------------------------------------------------------- /images/30.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/30.png -------------------------------------------------------------------------------- /images/31.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/31.png -------------------------------------------------------------------------------- /images/32.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/32.png -------------------------------------------------------------------------------- /images/33.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/33.png -------------------------------------------------------------------------------- /images/34.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/34.png -------------------------------------------------------------------------------- /images/35.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/35.png -------------------------------------------------------------------------------- /images/36.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/36.png -------------------------------------------------------------------------------- /images/37.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/37.png -------------------------------------------------------------------------------- /images/38.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/38.png -------------------------------------------------------------------------------- /images/39.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/39.png -------------------------------------------------------------------------------- /images/40.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/40.png -------------------------------------------------------------------------------- /images/41.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/41.png -------------------------------------------------------------------------------- /images/42.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/42.png -------------------------------------------------------------------------------- /images/43.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/43.png -------------------------------------------------------------------------------- /images/44.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/44.png -------------------------------------------------------------------------------- /images/45.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/45.png -------------------------------------------------------------------------------- /images/46.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/46.png -------------------------------------------------------------------------------- /images/47.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/47.png -------------------------------------------------------------------------------- /images/48.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/48.png -------------------------------------------------------------------------------- /images/49.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/49.png -------------------------------------------------------------------------------- /images/50.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/50.png -------------------------------------------------------------------------------- /images/51.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/51.png -------------------------------------------------------------------------------- /images/52.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/52.png -------------------------------------------------------------------------------- /images/53.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/53.png -------------------------------------------------------------------------------- /images/54.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/54.png -------------------------------------------------------------------------------- /images/55.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/55.png -------------------------------------------------------------------------------- /images/56.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/56.png -------------------------------------------------------------------------------- /images/57.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/57.png -------------------------------------------------------------------------------- /images/58.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/58.png -------------------------------------------------------------------------------- /images/59.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/59.png -------------------------------------------------------------------------------- /images/60.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/60.png -------------------------------------------------------------------------------- /images/61.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/61.png -------------------------------------------------------------------------------- /images/62.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/62.png -------------------------------------------------------------------------------- /images/63.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/63.png -------------------------------------------------------------------------------- /images/64.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/64.png -------------------------------------------------------------------------------- /images/65.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/65.png -------------------------------------------------------------------------------- /images/66.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/66.png -------------------------------------------------------------------------------- /images/67.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/67.png -------------------------------------------------------------------------------- /images/68.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/68.png -------------------------------------------------------------------------------- /images/69.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/69.png -------------------------------------------------------------------------------- /images/70.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/70.png -------------------------------------------------------------------------------- /images/71.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/71.png -------------------------------------------------------------------------------- /images/72.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/72.png -------------------------------------------------------------------------------- /images/73.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/73.png -------------------------------------------------------------------------------- /images/74.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/74.png -------------------------------------------------------------------------------- /images/75.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/75.png -------------------------------------------------------------------------------- /images/76.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/76.png -------------------------------------------------------------------------------- /images/77.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/77.png -------------------------------------------------------------------------------- /images/78.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/78.png -------------------------------------------------------------------------------- /images/79.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/79.png -------------------------------------------------------------------------------- /images/80.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/80.png -------------------------------------------------------------------------------- /images/81.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/81.png -------------------------------------------------------------------------------- /images/82.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/82.png -------------------------------------------------------------------------------- /images/83.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/83.png -------------------------------------------------------------------------------- /images/84.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/84.png -------------------------------------------------------------------------------- /images/85.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/85.png -------------------------------------------------------------------------------- /images/86.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/86.png -------------------------------------------------------------------------------- /images/87.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/87.png -------------------------------------------------------------------------------- /images/88.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/88.png -------------------------------------------------------------------------------- /images/89.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/89.png -------------------------------------------------------------------------------- /images/90.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/90.png -------------------------------------------------------------------------------- /images/91.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/91.png -------------------------------------------------------------------------------- /images/92.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/92.png -------------------------------------------------------------------------------- /images/93.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/93.png -------------------------------------------------------------------------------- /images/94.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/94.png -------------------------------------------------------------------------------- /images/95.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/95.png -------------------------------------------------------------------------------- /images/96.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/96.png -------------------------------------------------------------------------------- /images/97.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/97.png -------------------------------------------------------------------------------- /images/98.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/98.png -------------------------------------------------------------------------------- /images/99.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/99.png -------------------------------------------------------------------------------- /ml/cluster.md: -------------------------------------------------------------------------------- 1 | ![](/assets/clustermethod1.png)![](/assets/clustermethod2.png) 2 | 3 | -------------------------------------------------------------------------------- /assets/dlnet2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dlnet2.png -------------------------------------------------------------------------------- /assets/dlnet3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dlnet3.png -------------------------------------------------------------------------------- /assets/dlnet4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dlnet4.png -------------------------------------------------------------------------------- /assets/dlnet5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dlnet5.png -------------------------------------------------------------------------------- /assets/dlnet6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dlnet6.png -------------------------------------------------------------------------------- /assets/gif5.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/gif5.gif -------------------------------------------------------------------------------- /assets/ica2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/ica2.png -------------------------------------------------------------------------------- /assets/ica3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/ica3.png -------------------------------------------------------------------------------- /assets/mlpre1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre1.png -------------------------------------------------------------------------------- /assets/mlpre2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre2.png -------------------------------------------------------------------------------- /assets/mlpre3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre3.png -------------------------------------------------------------------------------- /assets/mlpre4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre4.png -------------------------------------------------------------------------------- /assets/mlpre5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre5.png -------------------------------------------------------------------------------- /assets/mlpre6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre6.png -------------------------------------------------------------------------------- /assets/mlpre7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre7.png -------------------------------------------------------------------------------- /assets/mlpre8.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre8.png -------------------------------------------------------------------------------- /assets/mlpre9.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre9.png -------------------------------------------------------------------------------- /assets/rl-td1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td1.png -------------------------------------------------------------------------------- /assets/rl-td2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td2.png -------------------------------------------------------------------------------- /assets/rl-td4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td4.png -------------------------------------------------------------------------------- /assets/rl-td48png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td48png -------------------------------------------------------------------------------- /assets/rl-td5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td5.png -------------------------------------------------------------------------------- /assets/rnnb1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rnnb1.png -------------------------------------------------------------------------------- /assets/rnnb2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rnnb2.png -------------------------------------------------------------------------------- /assets/rnnb3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rnnb3.png -------------------------------------------------------------------------------- /assets/ubc1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/ubc1.png -------------------------------------------------------------------------------- /assets/ucts.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/ucts.png -------------------------------------------------------------------------------- /assets/vi-em.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/vi-em.png -------------------------------------------------------------------------------- /images/svm/1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/svm/1.png -------------------------------------------------------------------------------- /assets/5_13_03.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/5_13_03.gif -------------------------------------------------------------------------------- /assets/5_13_06.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/5_13_06.gif -------------------------------------------------------------------------------- /assets/cnnconv1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/cnnconv1.png -------------------------------------------------------------------------------- /assets/cnnconv2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/cnnconv2.png -------------------------------------------------------------------------------- /assets/cnnconv3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/cnnconv3.png -------------------------------------------------------------------------------- /assets/cnnpool1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/cnnpool1.png -------------------------------------------------------------------------------- /assets/dlnet22.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dlnet22.png -------------------------------------------------------------------------------- /assets/dlnet23.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dlnet23.png -------------------------------------------------------------------------------- /assets/dlnet24.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dlnet24.png -------------------------------------------------------------------------------- /assets/dnnimg1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dnnimg1.png -------------------------------------------------------------------------------- /assets/dnnimg2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dnnimg2.png -------------------------------------------------------------------------------- /assets/dnnimg3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dnnimg3.png -------------------------------------------------------------------------------- /assets/dnnimg4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dnnimg4.png -------------------------------------------------------------------------------- /assets/dnnimg5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dnnimg5.png -------------------------------------------------------------------------------- /assets/dnnimg6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dnnimg6.png -------------------------------------------------------------------------------- /assets/dnnimg7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dnnimg7.png -------------------------------------------------------------------------------- /assets/encode4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/encode4.png -------------------------------------------------------------------------------- /assets/encoder1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/encoder1.png -------------------------------------------------------------------------------- /assets/encoder2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/encoder2.png -------------------------------------------------------------------------------- /assets/encoder3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/encoder3.png -------------------------------------------------------------------------------- /assets/encoder5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/encoder5.png -------------------------------------------------------------------------------- /assets/encoder7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/encoder7.png -------------------------------------------------------------------------------- /assets/ganvae1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/ganvae1.png -------------------------------------------------------------------------------- /assets/ganvae2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/ganvae2.png -------------------------------------------------------------------------------- /assets/ganvae3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/ganvae3.png -------------------------------------------------------------------------------- /assets/ganvae4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/ganvae4.png -------------------------------------------------------------------------------- /assets/ganvae5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/ganvae5.png -------------------------------------------------------------------------------- /assets/lg-kkt1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/lg-kkt1.png -------------------------------------------------------------------------------- /assets/lg-kkt2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/lg-kkt2.png -------------------------------------------------------------------------------- /assets/lg-kkt3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/lg-kkt3.png -------------------------------------------------------------------------------- /assets/lg-kkt4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/lg-kkt4.png -------------------------------------------------------------------------------- /assets/lg-kkt5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/lg-kkt5.png -------------------------------------------------------------------------------- /assets/lstmpci4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/lstmpci4.png -------------------------------------------------------------------------------- /assets/lstmpic1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/lstmpic1.png -------------------------------------------------------------------------------- /assets/lstmpic2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/lstmpic2.png -------------------------------------------------------------------------------- /assets/lstmpic3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/lstmpic3.png -------------------------------------------------------------------------------- /assets/lstmpic5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/lstmpic5.png -------------------------------------------------------------------------------- /assets/lstmpic6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/lstmpic6.png -------------------------------------------------------------------------------- /assets/lstmpic7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/lstmpic7.png -------------------------------------------------------------------------------- /assets/lstmpic8.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/lstmpic8.png -------------------------------------------------------------------------------- /assets/mdp-dp1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mdp-dp1.png -------------------------------------------------------------------------------- /assets/mdp-dp2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mdp-dp2.png -------------------------------------------------------------------------------- /assets/meituan1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/meituan1.png -------------------------------------------------------------------------------- /assets/meituan2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/meituan2.png -------------------------------------------------------------------------------- /assets/meituan3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/meituan3.png -------------------------------------------------------------------------------- /assets/meituan4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/meituan4.png -------------------------------------------------------------------------------- /assets/meituan5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/meituan5.png -------------------------------------------------------------------------------- /assets/meituan7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/meituan7.png -------------------------------------------------------------------------------- /assets/meituan8.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/meituan8.png -------------------------------------------------------------------------------- /assets/meituan9.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/meituan9.png -------------------------------------------------------------------------------- /assets/mlpre11.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre11.png -------------------------------------------------------------------------------- /assets/mlpre12.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre12.png -------------------------------------------------------------------------------- /assets/mlpre13.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre13.png -------------------------------------------------------------------------------- /assets/mlpre14.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre14.png -------------------------------------------------------------------------------- /assets/mlpre15.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre15.png -------------------------------------------------------------------------------- /assets/mlpre16.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre16.png -------------------------------------------------------------------------------- /assets/mlpre17.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlpre17.png -------------------------------------------------------------------------------- /assets/newton1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/newton1.png -------------------------------------------------------------------------------- /assets/restnet1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/restnet1.png -------------------------------------------------------------------------------- /assets/rl-td11.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td11.png -------------------------------------------------------------------------------- /assets/rl-td21.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td21.png -------------------------------------------------------------------------------- /assets/rl-td31.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td31.png -------------------------------------------------------------------------------- /assets/rl-td33.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td33.png -------------------------------------------------------------------------------- /assets/rl-td34.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td34.png -------------------------------------------------------------------------------- /assets/rl-td41.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td41.png -------------------------------------------------------------------------------- /assets/rl-td46.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td46.png -------------------------------------------------------------------------------- /assets/rl-td47.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td47.png -------------------------------------------------------------------------------- /assets/rl-td48.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td48.png -------------------------------------------------------------------------------- /assets/rl-td49.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td49.png -------------------------------------------------------------------------------- /assets/rl-td51.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td51.png -------------------------------------------------------------------------------- /assets/rl-td53.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td53.png -------------------------------------------------------------------------------- /assets/rl-td61.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td61.png -------------------------------------------------------------------------------- /assets/rl-td71.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td71.png -------------------------------------------------------------------------------- /assets/rl-td81.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td81.png -------------------------------------------------------------------------------- /assets/softmax1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/softmax1.png -------------------------------------------------------------------------------- /assets/softmax2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/softmax2.png -------------------------------------------------------------------------------- /assets/softmax3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/softmax3.png -------------------------------------------------------------------------------- /images/bp/0111.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/bp/0111.png -------------------------------------------------------------------------------- /images/gan/vae1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/gan/vae1.png -------------------------------------------------------------------------------- /images/gan/vae2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/gan/vae2.png -------------------------------------------------------------------------------- /images/gan/vae3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/gan/vae3.png -------------------------------------------------------------------------------- /images/gan/vae4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/gan/vae4.png -------------------------------------------------------------------------------- /images/gan/vae5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/gan/vae5.png -------------------------------------------------------------------------------- /images/gan/vae6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/gan/vae6.png -------------------------------------------------------------------------------- /images/gan/vae7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/gan/vae7.png -------------------------------------------------------------------------------- /images/gan/vae8.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/gan/vae8.png -------------------------------------------------------------------------------- /images/gan/vae9.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/gan/vae9.png -------------------------------------------------------------------------------- /images/overview.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/overview.png -------------------------------------------------------------------------------- /assets/cnnalexnet1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/cnnalexnet1.png -------------------------------------------------------------------------------- /assets/cnndensnet1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/cnndensnet1.png -------------------------------------------------------------------------------- /assets/cnndensnet2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/cnndensnet2.png -------------------------------------------------------------------------------- /assets/convetex1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/convetex1.png -------------------------------------------------------------------------------- /assets/convetex2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/convetex2.png -------------------------------------------------------------------------------- /assets/convetex3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/convetex3.png -------------------------------------------------------------------------------- /assets/dlintronet2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dlintronet2.png -------------------------------------------------------------------------------- /assets/gif5新文件 (1).gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/gif5新文件 (1).gif -------------------------------------------------------------------------------- /assets/googlenet1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/googlenet1.png -------------------------------------------------------------------------------- /assets/hmm-markv3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/hmm-markv3.png -------------------------------------------------------------------------------- /assets/hmm-markv4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/hmm-markv4.png -------------------------------------------------------------------------------- /assets/hmm-markv5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/hmm-markv5.png -------------------------------------------------------------------------------- /assets/intronet1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/intronet1.png -------------------------------------------------------------------------------- /assets/jiekadejuli.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/jiekadejuli.png -------------------------------------------------------------------------------- /assets/knncode100.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/knncode100.jpg -------------------------------------------------------------------------------- /assets/knncode100.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/knncode100.png -------------------------------------------------------------------------------- /assets/markv-mdp1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/markv-mdp1.png -------------------------------------------------------------------------------- /assets/markv-mdp11.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/markv-mdp11.png -------------------------------------------------------------------------------- /assets/markv-mdp2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/markv-mdp2.png -------------------------------------------------------------------------------- /assets/markv-mrp1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/markv-mrp1.png -------------------------------------------------------------------------------- /assets/meituan10.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/meituan10.png -------------------------------------------------------------------------------- /assets/meituan11.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/meituan11.png -------------------------------------------------------------------------------- /assets/meituan12.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/meituan12.png -------------------------------------------------------------------------------- /assets/meituan14.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/meituan14.png -------------------------------------------------------------------------------- /assets/meituan15.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/meituan15.png -------------------------------------------------------------------------------- /assets/meituan16.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/meituan16.png -------------------------------------------------------------------------------- /assets/multi-bt1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multi-bt1.png -------------------------------------------------------------------------------- /assets/multi-bt10.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multi-bt10.png -------------------------------------------------------------------------------- /assets/multi-bt11.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multi-bt11.png -------------------------------------------------------------------------------- /assets/multi-bt12.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multi-bt12.png -------------------------------------------------------------------------------- /assets/multi-bt2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multi-bt2.png -------------------------------------------------------------------------------- /assets/multi-bt3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multi-bt3.png -------------------------------------------------------------------------------- /assets/multi-bt5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multi-bt5.png -------------------------------------------------------------------------------- /assets/multi-bt7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multi-bt7.png -------------------------------------------------------------------------------- /assets/multi-bt8.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multi-bt8.png -------------------------------------------------------------------------------- /assets/multi-bt9.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multi-bt9.png -------------------------------------------------------------------------------- /assets/mywebchat.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mywebchat.png -------------------------------------------------------------------------------- /assets/prefixspan1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/prefixspan1.png -------------------------------------------------------------------------------- /assets/prefixspan2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/prefixspan2.png -------------------------------------------------------------------------------- /assets/prefixspan3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/prefixspan3.png -------------------------------------------------------------------------------- /assets/prefixspan4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/prefixspan4.png -------------------------------------------------------------------------------- /assets/prefixspan5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/prefixspan5.png -------------------------------------------------------------------------------- /assets/rbm-yuanli2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yuanli2.png -------------------------------------------------------------------------------- /assets/rbm-yuanli3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yuanli3.png -------------------------------------------------------------------------------- /assets/rbm-yuanli5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yuanli5.png -------------------------------------------------------------------------------- /assets/rbm-yuanli6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yuanli6.png -------------------------------------------------------------------------------- /assets/rbm-yuanli7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yuanli7.png -------------------------------------------------------------------------------- /assets/rbm-yuanli8.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yuanli8.png -------------------------------------------------------------------------------- /assets/rbm-yuanli9.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yuanli9.png -------------------------------------------------------------------------------- /assets/rbm-yunali1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yunali1.png -------------------------------------------------------------------------------- /assets/rbm-yunali4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yunali4.png -------------------------------------------------------------------------------- /assets/rl-td-error.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rl-td-error.png -------------------------------------------------------------------------------- /assets/runseqtask1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/runseqtask1.png -------------------------------------------------------------------------------- /assets/simplenet.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/simplenet.png -------------------------------------------------------------------------------- /assets/touyouhua10.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/touyouhua10.png -------------------------------------------------------------------------------- /assets/touyouhua11.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/touyouhua11.png -------------------------------------------------------------------------------- /assets/touyouhua2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/touyouhua2.png -------------------------------------------------------------------------------- /assets/touyouhua3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/touyouhua3.png -------------------------------------------------------------------------------- /assets/touyouhua4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/touyouhua4.png -------------------------------------------------------------------------------- /assets/touyouhua5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/touyouhua5.png -------------------------------------------------------------------------------- /assets/touyouhua6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/touyouhua6.png -------------------------------------------------------------------------------- /assets/touyouhua7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/touyouhua7.png -------------------------------------------------------------------------------- /assets/touyouhua8.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/touyouhua8.png -------------------------------------------------------------------------------- /assets/touyouhua9.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/touyouhua9.png -------------------------------------------------------------------------------- /assets/touyuhua1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/touyuhua1.png -------------------------------------------------------------------------------- /assets/xinxishang.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/xinxishang.png -------------------------------------------------------------------------------- /images/bp/01114321.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/bp/01114321.png -------------------------------------------------------------------------------- /images/gan/vae10.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/gan/vae10.png -------------------------------------------------------------------------------- /images/gan/vae11.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/gan/vae11.png -------------------------------------------------------------------------------- /assets/alirecomadn7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/alirecomadn7.png -------------------------------------------------------------------------------- /assets/alirecomamnd6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/alirecomamnd6.png -------------------------------------------------------------------------------- /assets/alirecommadn4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/alirecommadn4.png -------------------------------------------------------------------------------- /assets/alirecommand1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/alirecommand1.png -------------------------------------------------------------------------------- /assets/alirecommand2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/alirecommand2.png -------------------------------------------------------------------------------- /assets/alirecommand5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/alirecommand5.png -------------------------------------------------------------------------------- /assets/alirecoomand3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/alirecoomand3.png -------------------------------------------------------------------------------- /assets/commonf-rleu.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/commonf-rleu.png -------------------------------------------------------------------------------- /assets/commonf-tanh.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/commonf-tanh.png -------------------------------------------------------------------------------- /assets/deeplayer-bn1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplayer-bn1.png -------------------------------------------------------------------------------- /assets/deeplayer-bn2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplayer-bn2.png -------------------------------------------------------------------------------- /assets/deeplayer-bn3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplayer-bn3.png -------------------------------------------------------------------------------- /assets/makrv-bestvf.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/makrv-bestvf.png -------------------------------------------------------------------------------- /assets/markv-mdpvf1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/markv-mdpvf1.png -------------------------------------------------------------------------------- /assets/markv-policy1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/markv-policy1.png -------------------------------------------------------------------------------- /assets/markv-reward1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/markv-reward1.png -------------------------------------------------------------------------------- /assets/mento-calro2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mento-calro2.png -------------------------------------------------------------------------------- /assets/mento-carlo1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mento-carlo1.png -------------------------------------------------------------------------------- /assets/mento-carlo3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mento-carlo3.png -------------------------------------------------------------------------------- /assets/mento-carlo4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mento-carlo4.png -------------------------------------------------------------------------------- /assets/mento-carlo5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mento-carlo5.png -------------------------------------------------------------------------------- /assets/mento-carlo6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mento-carlo6.png -------------------------------------------------------------------------------- /assets/mento-carlo7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mento-carlo7.png -------------------------------------------------------------------------------- /assets/mento-carlo8.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mento-carlo8.png -------------------------------------------------------------------------------- /assets/rbm-yuanli10.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yuanli10.png -------------------------------------------------------------------------------- /assets/rbm-yuanli11.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yuanli11.png -------------------------------------------------------------------------------- /assets/rbm-yuanli21.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yuanli21.png -------------------------------------------------------------------------------- /assets/rbm-yuanli22.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yuanli22.png -------------------------------------------------------------------------------- /assets/rbm-yuanli23.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yuanli23.png -------------------------------------------------------------------------------- /assets/rbm-yuanli24.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yuanli24.png -------------------------------------------------------------------------------- /assets/rbm-yuanli25.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rbm-yuanli25.png -------------------------------------------------------------------------------- /assets/rnnstructure1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/rnnstructure1.png -------------------------------------------------------------------------------- /assets/voice-intro1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/voice-intro1.png -------------------------------------------------------------------------------- /assets/voice-intro2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/voice-intro2.png -------------------------------------------------------------------------------- /assets/voice-intro3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/voice-intro3.png -------------------------------------------------------------------------------- /assets/voice-intro4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/voice-intro4.png -------------------------------------------------------------------------------- /assets/voice-intro5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/voice-intro5.png -------------------------------------------------------------------------------- /assets/voice-intro6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/voice-intro6.png -------------------------------------------------------------------------------- /assets/voice-intro7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/voice-intro7.png -------------------------------------------------------------------------------- /images/dl/gan/clip1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/dl/gan/clip1.png -------------------------------------------------------------------------------- /images/dl/gan/clip2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/dl/gan/clip2.png -------------------------------------------------------------------------------- /images/dl/gan/import.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/dl/gan/import.png -------------------------------------------------------------------------------- /images/dl/gan/minmax.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/dl/gan/minmax.png -------------------------------------------------------------------------------- /images/dl/gan/table1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/dl/gan/table1.png -------------------------------------------------------------------------------- /assets/clustermethod1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/clustermethod1.png -------------------------------------------------------------------------------- /assets/clustermethod2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/clustermethod2.png -------------------------------------------------------------------------------- /assets/deeplearinggan4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplearinggan4.png -------------------------------------------------------------------------------- /assets/deeplearinggan5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplearinggan5.png -------------------------------------------------------------------------------- /assets/deeplearinggan6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplearinggan6.png -------------------------------------------------------------------------------- /assets/deeplearinggan8.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplearinggan8.png -------------------------------------------------------------------------------- /assets/deeplearninggan1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplearninggan1.png -------------------------------------------------------------------------------- /assets/deeplearninggan2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplearninggan2.png -------------------------------------------------------------------------------- /assets/deeplearninggan3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplearninggan3.png -------------------------------------------------------------------------------- /assets/deeplearninggan7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplearninggan7.png -------------------------------------------------------------------------------- /assets/deeplearninggan9.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplearninggan9.png -------------------------------------------------------------------------------- /assets/dl-cnn-for-nlp.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dl-cnn-for-nlp.png -------------------------------------------------------------------------------- /assets/hmm-mengtekaluo2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/hmm-mengtekaluo2.png -------------------------------------------------------------------------------- /assets/hmm-mentekaluo1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/hmm-mentekaluo1.png -------------------------------------------------------------------------------- /assets/label-recommand.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/label-recommand.png -------------------------------------------------------------------------------- /assets/layers-softmax.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/layers-softmax.png -------------------------------------------------------------------------------- /assets/logisticcode100.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/logisticcode100.jpg -------------------------------------------------------------------------------- /assets/markv-bellman1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/markv-bellman1.png -------------------------------------------------------------------------------- /assets/markv-bellman2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/markv-bellman2.png -------------------------------------------------------------------------------- /assets/markv-bellman3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/markv-bellman3.png -------------------------------------------------------------------------------- /assets/mlmodelmetrics10.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetrics10.png -------------------------------------------------------------------------------- /assets/mlmodelmetrics11.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetrics11.png -------------------------------------------------------------------------------- /assets/mlmodelmetrics12.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetrics12.png -------------------------------------------------------------------------------- /assets/mlmodelmetrics13.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetrics13.png -------------------------------------------------------------------------------- /assets/mlmodelmetrics14.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetrics14.png -------------------------------------------------------------------------------- /assets/mlmodelmetrics15.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetrics15.png -------------------------------------------------------------------------------- /assets/mlmodelmetrics16.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetrics16.png -------------------------------------------------------------------------------- /assets/mlmodelmetrics17.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetrics17.png -------------------------------------------------------------------------------- /assets/mlmodelmetrics2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetrics2.png -------------------------------------------------------------------------------- /assets/mlmodelmetrics3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetrics3.png -------------------------------------------------------------------------------- /assets/mlmodelmetrics32.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetrics32.png -------------------------------------------------------------------------------- /assets/mlmodelmetrics6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetrics6.png -------------------------------------------------------------------------------- /assets/mlmodelmetrics8.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetrics8.png -------------------------------------------------------------------------------- /assets/mlmodelmetrics9.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetrics9.png -------------------------------------------------------------------------------- /assets/mlmodemetrics1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodemetrics1.png -------------------------------------------------------------------------------- /assets/mlmodemetrics18.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodemetrics18.png -------------------------------------------------------------------------------- /assets/mlmodemetrics19.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodemetrics19.png -------------------------------------------------------------------------------- /assets/mlmodemetrics7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodemetrics7.png -------------------------------------------------------------------------------- /assets/multibandit-gba.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multibandit-gba.png -------------------------------------------------------------------------------- /assets/multibandit-gba2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multibandit-gba2.png -------------------------------------------------------------------------------- /assets/multilrcode100.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multilrcode100.jpg -------------------------------------------------------------------------------- /assets/mutilbandit-ucb.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mutilbandit-ucb.png -------------------------------------------------------------------------------- /assets/reinforcement-ql.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcement-ql.png -------------------------------------------------------------------------------- /assets/reinforcemnt-ql4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcemnt-ql4.png -------------------------------------------------------------------------------- /assets/runnbidirection1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/runnbidirection1.png -------------------------------------------------------------------------------- /assets/simplelrcode100.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/simplelrcode100.jpg -------------------------------------------------------------------------------- /images/cluster/table1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/cluster/table1.png -------------------------------------------------------------------------------- /images/cluster/table2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/cluster/table2.png -------------------------------------------------------------------------------- /images/cluster/table3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/cluster/table3.png -------------------------------------------------------------------------------- /images/cluster/table4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/cluster/table4.png -------------------------------------------------------------------------------- /images/dl/gan/gan3loss.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/dl/gan/gan3loss.png -------------------------------------------------------------------------------- /images/dl/gan/import2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/dl/gan/import2.png -------------------------------------------------------------------------------- /assets/dataprocesscode100.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dataprocesscode100.jpg -------------------------------------------------------------------------------- /assets/deeplayer-pooling1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplayer-pooling1.png -------------------------------------------------------------------------------- /assets/deeplearninggan10.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplearninggan10.png -------------------------------------------------------------------------------- /assets/deeplearninggan11.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplearninggan11.png -------------------------------------------------------------------------------- /assets/deeplearninggan14.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplearninggan14.png -------------------------------------------------------------------------------- /assets/deeplearninggan17.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplearninggan17.png -------------------------------------------------------------------------------- /assets/dl-popnet-resnet1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dl-popnet-resnet1.png -------------------------------------------------------------------------------- /assets/dl-popnet-resnet2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dl-popnet-resnet2.png -------------------------------------------------------------------------------- /assets/dl-popnet-resnet3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dl-popnet-resnet3.png -------------------------------------------------------------------------------- /assets/mlmodelmetircsd24.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetircsd24.png -------------------------------------------------------------------------------- /assets/mlmodelmetricsd22.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetricsd22.png -------------------------------------------------------------------------------- /assets/mlmodelmetricvs21.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlmodelmetricvs21.png -------------------------------------------------------------------------------- /assets/reinforcement-a3c.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcement-a3c.png -------------------------------------------------------------------------------- /assets/reinforcement-dqn1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcement-dqn1.png -------------------------------------------------------------------------------- /assets/reinforcement-dqn3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcement-dqn3.png -------------------------------------------------------------------------------- /assets/reinforcement-dqn4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcement-dqn4.png -------------------------------------------------------------------------------- /assets/reinforcement-dqn5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcement-dqn5.png -------------------------------------------------------------------------------- /assets/reinforcement-pg1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcement-pg1.png -------------------------------------------------------------------------------- /assets/reinforcement-ql2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcement-ql2.png -------------------------------------------------------------------------------- /assets/reinforcement-ql3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcement-ql3.png -------------------------------------------------------------------------------- /assets/reinforcemnt-dqn2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcemnt-dqn2.png -------------------------------------------------------------------------------- /images/bp/qwerqwreqerqer.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/bp/qwerqwreqerqer.png -------------------------------------------------------------------------------- /assets/decisiontreecode100.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/decisiontreecode100.jpg -------------------------------------------------------------------------------- /assets/deeplayer-embeding2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplayer-embeding2.png -------------------------------------------------------------------------------- /assets/deeplayer-embeding3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplayer-embeding3.png -------------------------------------------------------------------------------- /assets/deeplayerr-embemding1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplayerr-embemding1.png -------------------------------------------------------------------------------- /assets/deeplearinginggan13.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deeplearinginggan13.png -------------------------------------------------------------------------------- /assets/dl-poopnet-mobilenet1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dl-poopnet-mobilenet1.png -------------------------------------------------------------------------------- /assets/dl-popnet-mobilenet.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dl-popnet-mobilenet.png -------------------------------------------------------------------------------- /assets/dl-popnet-mobilenet10.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dl-popnet-mobilenet10.png -------------------------------------------------------------------------------- /assets/dl-popnet-mobilenet2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dl-popnet-mobilenet2.png -------------------------------------------------------------------------------- /assets/dl-popnet-mobilenet3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dl-popnet-mobilenet3.png -------------------------------------------------------------------------------- /assets/dl-popnet-mobilenet5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dl-popnet-mobilenet5.png -------------------------------------------------------------------------------- /assets/dl-popnet-mobilenet6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dl-popnet-mobilenet6.png -------------------------------------------------------------------------------- /assets/dl-popnet-mobilenet7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dl-popnet-mobilenet7.png -------------------------------------------------------------------------------- /assets/dl-popnet-mobilenet8.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dl-popnet-mobilenet8.png -------------------------------------------------------------------------------- /assets/dl-popnet-mobilenet9.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/dl-popnet-mobilenet9.png -------------------------------------------------------------------------------- /assets/integrate-learning1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/integrate-learning1.png -------------------------------------------------------------------------------- /assets/machinelearingdist2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/machinelearingdist2.png -------------------------------------------------------------------------------- /assets/machinelearningdist1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/machinelearningdist1.png -------------------------------------------------------------------------------- /assets/markv-transfer-matrix.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/markv-transfer-matrix.png -------------------------------------------------------------------------------- /assets/markv-valuefunction1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/markv-valuefunction1.png -------------------------------------------------------------------------------- /assets/markv-valuefunction2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/markv-valuefunction2.png -------------------------------------------------------------------------------- /assets/markv-vaulefunction3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/markv-vaulefunction3.png -------------------------------------------------------------------------------- /assets/multi-bootstraping2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multi-bootstraping2.png -------------------------------------------------------------------------------- /assets/multibandit-formula1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multibandit-formula1.png -------------------------------------------------------------------------------- /assets/multibandit-svgreward.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multibandit-svgreward.png -------------------------------------------------------------------------------- /assets/normal-expandvaiance.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/normal-expandvaiance.png -------------------------------------------------------------------------------- /assets/randomforestcode100.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/randomforestcode100.jpg -------------------------------------------------------------------------------- /assets/reinforcement-dqn21.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcement-dqn21.png -------------------------------------------------------------------------------- /assets/reinforcementlearing3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcementlearing3.png -------------------------------------------------------------------------------- /assets/reinforcementlearing5.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcementlearing5.png -------------------------------------------------------------------------------- /assets/reinforementlearing7.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforementlearing7.png -------------------------------------------------------------------------------- /images/bp/qewrqwreqrwerqwer.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/bp/qewrqwreqrwerqwer.png -------------------------------------------------------------------------------- /images/bp/weqrqwrcqwerqwer.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/bp/weqrqwrcqwerqwer.png -------------------------------------------------------------------------------- /images/cnn/weqrqwrcqwerqwer.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/cnn/weqrqwrcqwerqwer.png -------------------------------------------------------------------------------- /assets/reinforcementlearning1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcementlearning1.png -------------------------------------------------------------------------------- /assets/reinforcementlearning2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcementlearning2.png -------------------------------------------------------------------------------- /assets/reinforcementlearning4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcementlearning4.png -------------------------------------------------------------------------------- /assets/reinforcementlearning6.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcementlearning6.png -------------------------------------------------------------------------------- /assets/reinforcemnt-opeai-ppo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcemnt-opeai-ppo.png -------------------------------------------------------------------------------- /images/bp/qwerqrqwreqwerqwere.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/bp/qwerqrqwreqwerqwere.png -------------------------------------------------------------------------------- /assets/mlutibandit-rewardreduce.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/mlutibandit-rewardreduce.png -------------------------------------------------------------------------------- /assets/multi-steps-bootstraping1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/multi-steps-bootstraping1.png -------------------------------------------------------------------------------- /assets/reinforcement-deepmind-ppo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/reinforcement-deepmind-ppo.png -------------------------------------------------------------------------------- /styles/ebook.css: -------------------------------------------------------------------------------- 1 | .book .book-body .page-wrapper .page-inner section.normal table { 2 | width: 100%; 3 | overflow: auto; 4 | } -------------------------------------------------------------------------------- /styles/pdf.css: -------------------------------------------------------------------------------- 1 | .book .book-body .page-wrapper .page-inner section.normal table { 2 | width: 100%; 3 | overflow: auto; 4 | } -------------------------------------------------------------------------------- /images/bp/weqrqwrcqqwerqwrcwerqwer.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/bp/weqrqwrcqqwerqwrcwerqwer.png -------------------------------------------------------------------------------- /ml/pythonml/pythonji-qi-shu-xue-ku.md: -------------------------------------------------------------------------------- 1 | # numpy 2 | 3 | --- 4 | 5 | 6 | 7 | # scikit-learn 8 | 9 | --- 10 | 11 | 12 | 13 | -------------------------------------------------------------------------------- /styles/website.css: -------------------------------------------------------------------------------- 1 | .book .book-body .page-wrapper .page-inner section.normal table { 2 | width: 100%; 3 | overflow: auto; 4 | } -------------------------------------------------------------------------------- /assets/deterministic-value-iteration.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/assets/deterministic-value-iteration.png -------------------------------------------------------------------------------- /images/machine-learning-classification.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/machine-learning-classification.jpg -------------------------------------------------------------------------------- /images/overfitting/20151102210633783.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/overfitting/20151102210633783.png -------------------------------------------------------------------------------- /images/overfitting/20151102210852970.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/overfitting/20151102210852970.png -------------------------------------------------------------------------------- /images/overfitting/20151102211204226.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/overfitting/20151102211204226.png -------------------------------------------------------------------------------- /images/overfitting/20151102211214508.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/overfitting/20151102211214508.png -------------------------------------------------------------------------------- /images/overfitting/20151102211315525.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/overfitting/20151102211315525.png -------------------------------------------------------------------------------- /.gitattributes: -------------------------------------------------------------------------------- 1 | *.js linguist-language=python 2 | *.css linguist-language=python 3 | *.html linguist-language=python 4 | *.md linguist-language=python 5 | -------------------------------------------------------------------------------- /dl/popnet/resnet.md: -------------------------------------------------------------------------------- 1 | ![](/assets/dl-popnet-resnet1.png)![](/assets/dl-popnet-resnet2.png) 2 | 3 | 4 | 5 | ![](/assets/dl-popnet-resnet3.png) 6 | 7 | -------------------------------------------------------------------------------- /images/metrics/1042406-20161024154443875-2037260202.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/metrics/1042406-20161024154443875-2037260202.jpg -------------------------------------------------------------------------------- /images/metrics/1042406-20161024164359046-1869944207.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/geekan/Machine-Learning/HEAD/images/metrics/1042406-20161024164359046-1869944207.png -------------------------------------------------------------------------------- /ml/hmm/markv-mengtekluo.md: -------------------------------------------------------------------------------- 1 | ![](/assets/hmm-mentekaluo1.png)![](/assets/hmm-mengtekaluo2.png)![](/assets/hmm-markv3.png)![](/assets/hmm-markv4.png)![](/assets/hmm-markv5.png) 2 | 3 | -------------------------------------------------------------------------------- /math/analytic/lagelangri-kkt.md: -------------------------------------------------------------------------------- 1 | ![](/assets/lg-kkt1.png) 2 | ![](/assets/lg-kkt2.png) 3 | 4 | ![](/assets/lg-kkt3.png) 5 | 6 | ![](/assets/lg-kkt4.png) 7 | 8 | ![](/assets/lg-kkt5.png) 9 | 10 | # 11 | 12 | 13 | 14 | -------------------------------------------------------------------------------- /dl/layers/wrapper.md: -------------------------------------------------------------------------------- 1 | ## TimeDistributed包装器 {#timedistributed} 2 | 3 | --- 4 | 5 | 该包装器可以把一个层应用到输入的每一个时间步上 6 | 7 | 8 | 9 | ## Bidirectional包装器 {#bidirectional} 10 | 11 | --- 12 | 13 | 双向RNN包装器 14 | 15 | 16 | 17 | 18 | 19 | -------------------------------------------------------------------------------- /liu-xing-wang-luo-jie-gou.md: -------------------------------------------------------------------------------- 1 | alexnet![](/assets/cnnalexnet1.png) 2 | 3 | google net![](/assets/googlenet1.png) 4 | 5 | resnet 6 | 7 | ![](/assets/restnet1.png) 8 | 9 | densnet 10 | 11 | ![](/assets/cnndensnet1.png)![](/assets/cnndensnet2.png) 12 | 13 | -------------------------------------------------------------------------------- /datasets/Data.csv: -------------------------------------------------------------------------------- 1 | Country,Age,Salary,Purchased 2 | France,44,72000,No 3 | Spain,27,48000,Yes 4 | Germany,30,54000,No 5 | Spain,38,61000,No 6 | Germany,40,,Yes 7 | France,35,58000,Yes 8 | Spain,,52000,No 9 | France,48,79000,Yes 10 | Germany,50,83000,No 11 | France,37,67000,Yes -------------------------------------------------------------------------------- /dl/layers/layers.md: -------------------------------------------------------------------------------- 1 | 深度学习的总体来讲分三层，输入层，隐藏层和输出层。如下图： 2 | 3 | ![](http://images2015.cnblogs.com/blog/1042406/201702/1042406-20170220122136538-2002639053.png) 4 | 5 | 但是中间的隐藏层可以是多层，所以叫深度神经网络，中间的隐藏层可以有多种形式，就构成了各种不同的神经网络模型。这部分主要介绍各种常见的神经网络层。在熟悉这些常见的层后，一个神经网络其实就是各种不同层的组合。后边介绍主要基于keras的文档进行组织介绍。 6 | 7 | -------------------------------------------------------------------------------- /math/linear-matrix/lapack.md: -------------------------------------------------------------------------------- 1 | LAPACK是由美国国家科学基金等资助开发的著名公开软件。LAPACK包含了求解科学与工程计算中最常见的数值线性代数问题，如求解[线性方程组](https://baike.baidu.com/item/线性方程组)、线性最小二乘问题、[特征值](https://baike.baidu.com/item/特征值)问题和[奇异值](https://baike.baidu.com/item/奇异值)问题等。 2 | 3 | Numpy linalg底层调用LAPACK. 4 | 5 | 6 | 7 | BLAS 8 | 9 | -------------------------------------------------------------------------------- /ml/clean-feature/outlier-detect.md: -------------------------------------------------------------------------------- 1 | # 异常值检测 2 | 3 | 1，极值分析 4 | 5 | 通过scatterplots,histogramas, box和whisker plot分析极值。 6 | 7 | 查看样本分布（假设高斯分布），去距离1/4和3/4值2-3倍标准差数值的样本。 8 | 9 | 2，临近方法 10 | 11 | 基于k-means分析样本质心，去掉离质心特别远的样本。 12 | 13 | 3，投影方法 14 | 15 | 通过PCA，SOM，sammon mapping去掉不重要特征。 16 | 17 | -------------------------------------------------------------------------------- /ml/decisiontree.md: -------------------------------------------------------------------------------- 1 | 决策树\(Decision Tree）是在已知各种情况发生概率的基础上，通过构成决策树来求取净现值的期望值大于等于零的概率，评价项目风险，判断其可行性的决策分析方法，是直观运用概率分析的一种图解法。由于这种决策分支画成图形很像一棵树的枝干，故称决策树。在机器学习中，决策树是一个预测模型，他代表的是对象属性与对象值之间的一种映射关系。Entropy = 系统的凌乱程度，使用算法ID3C4.5,和C5.0生成树算法使用熵。这一度量是基于信息学理论中熵的概念。 2 | 3 | 决策树是一种树形结构，其中每个内部节点表示一个属性上的测试，每个分支代表一个测试输出，每个叶节点代表一种类别。 4 | 5 | -------------------------------------------------------------------------------- /ml/crf/crf.md: -------------------------------------------------------------------------------- 1 | CRF\(Conditional Random Field\) 条件随机场是近几年自然语言处理领域常用的算法之一，常用于句法分析、命名实体识别、词性标注等。在我看来，CRF就像一个反向的隐马尔可夫模型\(HMM\)，两者都是用了马尔科夫链作为隐含变量的概率转移模型，只不过HMM使用隐含变量生成可观测状态，其生成概率由标注集统计得到，是一个生成模型；而CRF反过来通过可观测状态判别隐含变量，其概率亦通过标注集统计得来，是一个判别模型。由于两者模型主干相同，其能够应用的领域往往是重叠的，但在命名实体、句法分析等领域CRF更胜一筹。当然你并不必须学习HMM才能读懂CRF，但通常来说如果做自然语言处理，这两个模型应该都有了解。 2 | 3 | -------------------------------------------------------------------------------- /datasets/studentscores.csv: -------------------------------------------------------------------------------- 1 | Hours,Scores 2 | 2.5,21 3 | 5.1,47 4 | 3.2,27 5 | 8.5,75 6 | 3.5,30 7 | 1.5,20 8 | 9.2,88 9 | 5.5,60 10 | 8.3,81 11 | 2.7,25 12 | 7.7,85 13 | 5.9,62 14 | 4.5,41 15 | 3.3,42 16 | 1.1,17 17 | 8.9,95 18 | 2.5,30 19 | 1.9,24 20 | 6.1,67 21 | 7.4,69 22 | 2.7,30 23 | 4.8,54 24 | 3.8,35 25 | 6.9,76 26 | 7.8,86 27 | -------------------------------------------------------------------------------- /dl/popnet/mobilenet.md: -------------------------------------------------------------------------------- 1 | ![](/assets/dl-poopnet-mobilenet1.png)![](/assets/dl-popnet-mobilenet2.png)![](/assets/dl-popnet-mobilenet3.png)![](/assets/dl-popnet-mobilenet.png)![](/assets/dl-popnet-mobilenet5.png)![](/assets/dl-popnet-mobilenet6.png)![](/assets/dl-popnet-mobilenet7.png)![](/assets/dl-popnet-mobilenet8.png)![](/assets/dl-popnet-mobilenet9.png)![](/assets/dl-popnet-mobilenet10.png) 2 | 3 | -------------------------------------------------------------------------------- /ml/evolution/yichuansuanfa.md: -------------------------------------------------------------------------------- 1 | **在程序里生宝宝, 杀死不乖的宝宝, 让乖宝宝继续生宝宝** 2 | 3 | 所有的遗传算法 \(Genetic Algorithm\), 后面都简称 GA, 我们都需要一个评估好坏的方程, 这个方程通常被称为 fitness 4 | 5 | 在 GA 中有基因, 为了方便, 我们直接就称为`DNA`吧. GA 中第二重要的就是这`DNA`了, 如何编码和解码`DNA`, 就是你使用 GA 首先要想到的问题. 传统的 GA 中,`DNA`我们能用一串二进制来表示 6 | 7 | 进化分三步: 8 | 9 | * 适者生存 \(selection\) 10 | * DNA 交叉配对 \(crossover\) 11 | * DNA 变异 \(mutation\) 12 | 13 | 14 | 15 | -------------------------------------------------------------------------------- /dl/encoder/sparse-autoencoder.md: -------------------------------------------------------------------------------- 1 | **Sparse AutoEncoder稀疏自动编码器：** 2 | 3 | 当然，我们还可以继续加上一些约束条件得到新的Deep Learning方法，如：如果在AutoEncoder的基础上加上L1的Regularity限制（L1主要是约束每一层中的节点中大部分都要为0，只有少数不为0，这就是Sparse名字的来源），我们就可以得到Sparse AutoEncoder法。 4 | 5 | ![](http://img.my.csdn.net/uploads/201304/09/1365439878_3585.jpg) 6 | 7 | 如上图，其实就是限制每次得到的表达code尽量稀疏。因为稀疏的表达往往比其他的表达要有效（人脑好像也是这样的，某个输入只是刺激某些神经元，其他的大部分的神经元是受到抑制的）。 8 | 9 | -------------------------------------------------------------------------------- /book.json: -------------------------------------------------------------------------------- 1 | { 2 | "title":"机器学习", 3 | "author": "shunliz", 4 | "description":"机器学习原理详解", 5 | "language": "zh", 6 | "plugins": ["katex", 7 | "image-captions" 8 | ], 9 | "pluginsConfig": { 10 | "image-captions": { 11 | "attributes": { 12 | "width": "600" 13 | }, 14 | "caption": "图 _PAGE_LEVEL_._PAGE_IMAGE_NUMBER_ - _CAPTION_" 15 | } 16 | } 17 | } -------------------------------------------------------------------------------- /dl/layers/lcnn.md: -------------------------------------------------------------------------------- 1 | ## LocallyConnected1D层 {#locallyconnected1d} 2 | 3 | --- 4 | 5 | `LocallyConnected1D`层与`Conv1D`工作方式类似，唯一的区别是不进行权值共享。即施加在不同输入位置的滤波器是不一样的。 6 | 7 | 8 | 9 | ## LocallyConnected2D层 {#locallyconnected2d} 10 | 11 | --- 12 | 13 | `LocallyConnected2D`层与`Convolution2D`工作方式类似，唯一的区别是不进行权值共享。即施加在不同输入patch的滤波器是不一样的，当使用该层作为模型首层时，需要提供参数`input_dim`或`input_shape`参数。参数含义参考`Convolution2D`。 14 | 15 | 16 | 17 | 18 | 19 | -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | # Node rules: 2 | ## Grunt intermediate storage (http://gruntjs.com/creating-plugins#storing-task-files) 3 | .grunt 4 | 5 | ## Dependency directory 6 | ## Commenting this out is preferred by some people, see 7 | ## https://docs.npmjs.com/misc/faq#should-i-check-my-node_modules-folder-into-git 8 | node_modules 9 | 10 | # Book build output 11 | _book 12 | 13 | # eBook build output 14 | *.epub 15 | *.mobi 16 | *.pdf 17 | -------------------------------------------------------------------------------- /dl/rbm/rbm-yuanli.md: -------------------------------------------------------------------------------- 1 | ![](/assets/rbm-yunali1.png)![](/assets/rbm-yuanli2.png)![](/assets/rbm-yuanli3.png)![](/assets/rbm-yunali4.png)![](/assets/rbm-yuanli5.png)![](/assets/rbm-yuanli6.png)![](/assets/rbm-yuanli7.png)![](/assets/rbm-yuanli8.png)![](/assets/rbm-yuanli9.png)![](/assets/rbm-yuanli10.png)![](/assets/rbm-yuanli11.png)![](/assets/rbm-yuanli21.png)![](/assets/rbm-yuanli22.png)![](/assets/rbm-yuanli23.png)![](/assets/rbm-yuanli24.png)![](/assets/rbm-yuanli25.png) 2 | 3 | -------------------------------------------------------------------------------- /dl/layers/merge.md: -------------------------------------------------------------------------------- 1 | # Merge层 {#merge} 2 | 3 | Merge层提供了一系列用于融合两个层或两个张量的层对象和方法。以大写首字母开头的是Layer类，以小写字母开头的是张量的函数。小写字母开头的张量函数在内部实际上是调用了大写字母开头的层。 4 | 5 | ## Add {#add} 6 | 7 | ## Multiply {#multiply} 8 | 9 | ## Maximum {#maximum} 10 | 11 | ## Concatenate {#concatenate} 12 | 13 | ## Dot {#dot} 14 | 15 | ## add {#add_1} 16 | 17 | ## multiply {#multiply_1} 18 | 19 | ## maximum {#maximum_1} 20 | 21 | ## concatenate {#concatenate_1} 22 | 23 | ## dot {#dot_1} 24 | 25 | 26 | 27 | 28 | 29 | -------------------------------------------------------------------------------- /dl/reinforcement/a3csuan-fa.md: -------------------------------------------------------------------------------- 1 | `A3C`的算法实际上就是将`Actor-Critic`放在了多个线程中进行同步训练. 可以想象成几个人同时在玩一样的游戏, 而他们玩游戏的经验都会同步上传到一个中央大脑. 然后他们又从中央大脑中获取最新的玩游戏方法. 2 | 3 | **这样, 对于这几个人, 他们的好处是:**中央大脑汇集了所有人的经验, 是最会玩游戏的一个, 他们能时不时获取到中央大脑的必杀招, 用在自己的场景中. 4 | 5 | **对于中央大脑的好处是:**中央大脑最怕一个人的连续性更新, 不只基于一个人推送更新这种方式能打消这种连续性. 使中央大脑不必有用像`DQN`,`DDPG`那样的记忆库也能很好的更新. 6 | 7 | ![](/assets/reinforcement-a3c.png)为了达到这个目的, 我们要有两套体系, 可以看作中央大脑拥有`global net`和他的参数, 每位玩家有一个`global net` 8 | 9 | 的副本`local net`, 可以定时向`global net`推送更新, 然后定时从`global net`那获取综合版的更新. 10 | 11 | -------------------------------------------------------------------------------- /dl/layers/nosie.md: -------------------------------------------------------------------------------- 1 | ## GaussianNoise层 {#gaussiannoise} 2 | 3 | --- 4 | 5 | 为数据施加0均值，标准差为`stddev`的加性高斯噪声。该层在克服过拟合时比较有用，你可以将它看作是随机的数据提升。高斯噪声是需要对输入数据进行破坏时的自然选择。 6 | 7 | 因为这是一个起正则化作用的层，该层只在训练时才有效。 8 | 9 | 10 | 11 | ## GaussianDropout层 {#gaussiandropout} 12 | 13 | --- 14 | 15 | 为层的输入施加以1为均值，标准差为`sqrt(rate/(1-rate)`的乘性高斯噪声 16 | 17 | 因为这是一个起正则化作用的层，该层只在训练时才有效。 18 | 19 | 20 | 21 | ## AlphaDropout {#alphadropout} 22 | 23 | --- 24 | 25 | 对输入施加Alpha Dropout 26 | 27 | Alpha Dropout是一种保持输入均值和方差不变的Dropout，该层的作用是即使在dropout时也保持数据的自规范性。 通过随机对负的饱和值进行激活，Alphe Drpout与selu激活函数配合较好。 28 | 29 | 30 | 31 | 32 | 33 | -------------------------------------------------------------------------------- /dl/layers/core.md: -------------------------------------------------------------------------------- 1 | 全连接层 2 | 3 | 全连接层实现了output = activation\(dot\(input, kernel\) + bias\) 4 | 5 | 全连接的的神经网络层。 6 | 7 | --- 8 | 9 | 激活层 10 | 11 | 对输出使用激活函数。单一层对神经网络输出施加一个激活函数。 12 | 13 | --- 14 | 15 | Dropout层 16 | 17 | 对输入数据进行适当的丢弃 18 | 19 | --- 20 | 21 | Flatten 22 | 23 | 对输入进行Flatten，不影响batch 大小 24 | 25 | --- 26 | 27 | Reshape 28 | 29 | 对输出重新定义唯独。 30 | 31 | --- 32 | 33 | Permute 34 | 35 | 根据一定的模型改变输入序列 36 | 37 | --- 38 | 39 | RepeatVector 40 | 41 | 重复输入n次 42 | 43 | --- 44 | 45 | Lambda 46 | 47 | 自定义层，嵌入自定义的表达式。 48 | 49 | --- 50 | 51 | softmax层 52 | 53 | --- 54 | 55 | $$softmax(x)_i=\frac {exp(x_i)}{\sum_{j}exp(x_j)} $$ 56 | 57 | tensorflow中定义的基本层，通常是对某一个NN的输出进行转化，转化成各个输出的概率。 58 | 59 | ![](/assets/layers-softmax.png) 60 | 61 | -------------------------------------------------------------------------------- /dl/rnn/bidirectional-rnns.md: -------------------------------------------------------------------------------- 1 | There have been many advancements with RNNs \([attention](https://www.oreilly.com/ideas/interpretability-via-attentional-and-memory-based-interfaces-using-tensorflow), Quasi RNNs, etc.\) that we will cover in later lessons but one of the basic and widely used ones are bidirectional RNNs \(Bi-RNNs\). The motivation behind bidirectional RNNs is to process an input sequence by both directions. Accounting for context from both sides can aid in performance when the entire input sequence is known at time of inference. A common application of Bi-RNNs is in translation where it's advantageous to look at an entire sentence from both sides when translating to another language \(ie. Japanese → English\). 2 | 3 | ![](/assets/runnbidirection1.png) 4 | 5 | 6 | 7 | -------------------------------------------------------------------------------- /ml/regression/regression.md: -------------------------------------------------------------------------------- 1 | 机器学习中的回归问题属于有监督学习的范畴。回归问题的目标是给定D维输入变量x，并且每一个输入矢量x都有对应的值y，要求对于新来的数据预测它对应的连续的目标值t。比如下面这个例子：假设我们有一个包含47个房子的面积和价格的数据集如下： 2 | 3 | ![](http://images.cnitblog.com/blog/392228/201410/291919410655805.jpg) 4 | 5 | 我们可以在Matlab中画出来这组数据集，如下： 6 | 7 | ![](http://images.cnitblog.com/blog/392228/201410/291921072538240.jpg) 8 | 9 | 　　看到画出来的点，是不是有点像一条直线？我们可以用一条曲线去尽量拟合这些数据点，那么对于新来的输入，我么就可以将拟合的曲线上返回对应的点从而达到预测的目的。如果要预测的值是连续的比如上述的房价，那么就属于回归问题；如果要预测的值是离散的即一个个标签，那么就属于分类问题。这个学习处理过程如下图所示： 10 | 11 | ![](http://images.cnitblog.com/blog/392228/201410/291925279255104.jpg) 12 | 13 | 　　上述学习过程中的常用术语：包含房子面积和价格的数据集称为**训练集training set；**输入变量x（本例中为面积）为**特征features；**输出的预测值y（本例中为房价）为**目标值target；**拟合的曲线，一般表示为y = h\(x\)，称为**假设模型hypothesis；**训练集的条目数称为**特征的维数**，本例为47。 14 | 15 | -------------------------------------------------------------------------------- /dl/word2vec/word2vec.md: -------------------------------------------------------------------------------- 1 | ## Word2Vec简介 2 | 3 | --- 4 | 5 | Word2vec是一个用于处理文本的双层神经网络。它的输入是文本语料，输出则是一组向量：该语料中词语的特征向量。虽然Word2vec并不是深度神经网络，但它可以将文本转换为深度神经网络能够理解的数值形式。 6 | 7 | Word2vec的应用不止于解析自然语句。它还可以用于基因组、代码、点赞、播放列表、社交媒体图像等其他语言或符号序列，同样能够有效识别其中存在的模式。 8 | 9 | 为什么呢？因为这些数据都是与词语相似的离散状态，而我们的目的只是求取这些状态之间的转移概率，即它们共同出现的可能性。所以gene2vec、like2vec和follower2vec都是可行的。 10 | 11 | Word2vec的目的和功用是在向量空间内将词的向量按相似性进行分组。它能够识别出数学上的相似性。Word2vec能生成向量，以分布式的数值形式来表示词的上下文等特征。而这一过程无需人工干预。 12 | 13 | 给出足够的数据、用法和上下文，Word2vec就能根据过去经验对词的意义进行高度准确的预测。这样的预测结果可以用于建立一个词与其他词之间的联系（例如，“男人”和“男孩”的关系与“女人”和“女孩”的关系相同），或者可以将文档聚类并按主题分类。而这些聚类结果是搜索、情感分析和推荐算法的基础，广泛应用于科研、调查取证、电子商务、客户关系管理等领域。 14 | 15 | Word2vec神经网络的输出是一个词汇表，其中每个词都有一个对应的向量，可以将这些向量输入深度学习网络，也可以只是通过查询这些向量来识别词之间的关系。 16 | 17 | Word2vec衡量词的余弦相似性，无相似性表示为90度角，而相似度为1的完全相似则表示为0度角，即完全重合 18 | 19 | -------------------------------------------------------------------------------- /CONTRIBUTING.md: -------------------------------------------------------------------------------- 1 | ## 如何贡献项目 2 | 3 | * 在 GitHub 上`fork`到自己的仓库，如yourrepo`/MachineLearning`，然后`clone`到本地，并设置用户信息。 4 | 5 | ``` 6 | $ git clone git@github.com/shunliz/MachineLearning.git 7 | $ 8 | cd MachineLearning 9 | $ git config user.name "yourname" 10 | 11 | $ git config user.email "your email" 12 | ``` 13 | 14 | * 修改代码后提交，并推送到自己的仓库。 15 | 16 | ``` 17 | $ #do some change on the content 18 | 19 | $ git commit -am "Fix issue #1: change helo to hello" 20 | 21 | $ git push 22 | ``` 23 | 24 | * 在 GitHub 网站上提交 pull request。 25 | 26 | * 定期使用项目仓库内容更新自己仓库内容。 27 | 28 | ``` 29 | $ git remote add upstream https://github.com/shunliz/MachineLearning 30 | $ git fetch upstream 31 | $ git checkout master 32 | $ git rebase upstream/master 33 | $ git push -f origin master 34 | ``` 35 | 36 | 37 | 38 | -------------------------------------------------------------------------------- /ml/clean-feature/one-hot.md: -------------------------------------------------------------------------------- 1 | One-Hot 编码，又称一位有效编码，其方法是使用N位状态寄存器来对N个状态进行编码，每个状态都由他独立的寄存器位，并且在任意时候，其中只有一位有效。 2 | 3 | 4 | 例如： 5 | 6 | > 自然状态码为：000,001,010,011,100,101 7 | > 8 | > 9 | > 独热编码为：000001,000010,000100,001000,010000,100000 10 | 11 | 可以这样理解，对于每一个特征，如果它有m个可能值，那么经过独热编码后，就变成了m个二元特征。并且，这些特征互斥，每次只有一个激活。因此，数据会变成稀疏的。 12 | 13 | 14 | 这样做的好处主要有： 15 | 16 | 1. 解决了分类器不好处理属性数据的问题 17 | 18 | 2. 在一定程度上也起到了扩充特征的作用 19 | 20 | 举例 21 | 22 | 23 | 我们基于[Python](http://lib.csdn.net/base/python)和Scikit-learn写一个简单的例子： 24 | 25 | > from sklearn import preprocessing 26 | > 27 | > 28 | > enc = preprocessing.OneHotEncoder\(\) 29 | > 30 | > enc.fit\(\[\[0, 0, 3\], \[1, 1, 0\], \[0, 2, 1\], \[1, 0, 2\]\]\) 31 | > 32 | > enc.transform\(\[\[0, 1, 3\]\]\).toarray\(\) 33 | 34 | 输出结果： 35 | 36 | 37 | > array\(\[\[ 1.,  0.,  0.,  1.,  0.,  0.,  0.,  0.,  1.\]\]\) 38 | 39 | 40 | 41 | -------------------------------------------------------------------------------- /dl/reinforcement/ppo.md: -------------------------------------------------------------------------------- 1 | **PPO: OpenAI 提出的一种解决 Policy Gradient 不好确定 Learning rate \(或者 Step size\) 的问题. 因为如果 step size 过大, 学出来的 Policy 会一直乱动, 不会收敛, 但如果 Step Size 太小, 对于完成训练, 我们会等到绝望. PPO 利用 New Policy 和 Old Policy 的比例, 限制了 New Policy 的更新幅度, 让 Policy Gradient 对稍微大点的 Step size 不那么敏感.** 2 | 3 | ![](/assets/reinforcemnt-opeai-ppo.png) 4 | 5 | ![](/assets/reinforcement-deepmind-ppo.png) 6 | 7 | 总的来说 PPO 是一套 Actor-Critic 结构, Actor 想**最大化**`J_PPO`, Critic 想**最小化**`L_BL`.Critic 的 loss 好说, 就是减小 TD error. 而 Actor 的就是在 old Policy 上根据 Advantage \(TD error\) 修改 new Policy, advantage 大的时候, 修改幅度大, 让 new Policy 更可能发生. 而且他们附加了一个 KL Penalty \(惩罚项, 不懂的同学搜一下 KL divergence\), 简单来说, 如果 new Policy 和 old Policy 差太多, 那 KL divergence 也越大, 我们不希望 new Policy 比 old Policy 差太多, 如果会差太多, 就相当于用了一个大的 Learning rate, 这样是不好的, 难收敛. 8 | 9 | 10 | 11 | DPPO 算法 12 | 13 | 使用多个线程 \(workers\) 平行在不同的环境中收集数据,运行PPO 14 | 15 | 16 | 17 | 18 | 19 | -------------------------------------------------------------------------------- /dl/rnn/rnn.md: -------------------------------------------------------------------------------- 1 | RNN的思想是利用序列信息。在经典的神经网络中，我们认为所有的输入和输出是相互独立的。 但是对于许多其他任务这个想法是很不好的。如果你需要预测一个句子中的下一个字，知道前边的字会是很有帮助的。理论上RNN可以利用任意长度的序列的信息，但是实际只能处理很有限的前几步信息。下边是一个典型的RNN图示： 2 | 3 | ![](/assets/rnnb1.png) 4 | 5 | 上图是一个展开的RNN，也就是说如果我们要处理一个含有5个字的序列，这个网络就要展开成5层的神经网络，一层处理一个字。图中的各个符号解释如下： 6 | 7 | $$x_t$$是第t步的输入，$$x_1$$可能是一个one-hot编码的的字。 8 | 9 | $$s_t$$是第t步的隐藏状态，是网络的记忆功能。$$s_t$$是通过前边的隐藏状态和当前输入步计算得出：$$st=f(Uxt+Ws_{t-1})$$, f通常是tanh或者ReLU 10 | 11 | $$o_t$$是第t步的输出，$$ot=softmax(Vs_t)$$ 12 | 13 | RNN的数学描述： 14 | 15 | $$s_t=tanh(Uxt+Ws_{t-1})$$ 16 | 17 | $$ot=softmax(Vs_t)$$ 18 | 19 | $$L(y,o)=-1/N\sum_{n}^{} y_nlogo_n$$ 20 | 21 | RNN在很多NLP任务中表现的非常出色。 22 | 23 | ### RNN通用来生成文本 24 | 25 | 通过给定一段文本，通过一定的训练，可以生成相似的文本。 26 | 27 | ### 机器翻译 28 | 29 | ![](/assets/rnnb2.png)根据原语言的输入序列，预测目标语言的序列达到翻译的目的。 30 | 31 | ### 语音识别 32 | 33 | 输入一段音频序列，预测出一个序列的音频。 34 | 35 | ### 生成文本描述 36 | 37 | ![](/assets/rnnb3.png)和CNN一块，CNN识别图像内容，RNN生成图像的描述。 38 | 39 | -------------------------------------------------------------------------------- /ml/pythonml/mo-xing-xuan-ze.md: -------------------------------------------------------------------------------- 1 | 经常地，对一堆数据进行建模的时候，特别是分类和回归模型，我们有很多的变量可供使用，选择不同的变量组合可以得到不同的模型，例如我们有5个变量，2的5次方，我们将有32个变量组合，可以训练出32个模型。但是哪个模型更加的好呢？目前常用有如下方法： 2 | 3 | AIC=-2 ln\(L\) + 2 k  中文名字：赤池信息量 akaike information criterion 4 | 5 | BIC=-2 ln\(L\) + ln\(n\)\*k 中文名字：贝叶斯信息量 bayesian information criterion 6 | 7 | HQ=-2 ln\(L\) + ln\(ln\(n\)\)\*k  hannan-quinn criterion 8 | 9 | 其中L是在该模型下的最大似然，n是数据数量，k是模型的变量个数。 10 | 11 | 选择最优模型的指导思想是从两个方面去考察：一个是似然函数最大化，另一个是模型中的未知参数个数最小化。似然函数值越大说明模型拟合的效果越好，但是我们不能单纯地以拟合精度来衡量模型的优劣，这样回导致模型中未知参数越来越多，模型变得越来越复杂，会造成过拟合。所以一个好的模型应该是拟合精度和未知参数个数的综合最优化配置。 12 | 13 | 14 | 15 | 注意这些规则只是刻画了用某个模型之后相对“真实模型”的信息损失【因为不知道真正的模型是什么样子，所以训练得到的所有模型都只是真实模型的一个近似模型】，所以用这些规则不能说明某个模型的精确度，即三个模型A, B, C，在通过这些规则计算后，我们知道B模型是三个模型中最好的，但是不能保证B这个模型就能够很好地刻画数据，因为很有可能这三个模型都是非常糟糕的，B只是烂苹果中的相对好的苹果而已。 16 | 17 | 这些规则理论上是比较漂亮的，但是实际在模型选择中应用起来还是有些困难的，例如上面我们说了5个变量就有32个变量组合，如果是10个变量呢？2的10次方，我们不可能对所有这些模型进行一一验证AIC, BIC，HQ规则来选择模型，工作量太大。 18 | 19 | 20 | 21 | -------------------------------------------------------------------------------- /math/analytic/tuyouhua.md: -------------------------------------------------------------------------------- 1 | Jensen不等式 2 | 3 | 如果$$f: \omega->R$$是一个函数，则对于任何$$[{ x_i \in \Omega }]^{n}_{i=1}$$以及凸组合$$\sum_{i=1}^{n} w_ix_i$$都有 4 | 5 | $$\sum_{i=1}^{n} w_if(x_i)>=f(\sum_{i=1}^{n} w_ix_i)$$ 6 | 7 | --- 8 | 9 | 拉格朗日对偶函数 10 | 11 | $$L(x,\lambda ,v) = f_0(x)+\sum_{i=1}^{m}\lambda _if_i(x)+\sum_{i=1}^{P}v_ih_i(x)$$ 12 | 13 | 根据拉个朗日函数，我们定义i拉格朗日对偶函数$$g(\lambda,v):R^{m+p}->R$$ 14 | 15 | $$x = y$$$$x = g(\lambda,v)=inf_{x \in D} L(x, \lambda, v)=inf_{x \in D} f_0(x)+\sum_{i=1}^{m}\lambda _if_i(x)+\sum_{i=1}^{P}v_ih_i(x)$$ 16 | 17 | --- 18 | 19 | 对偶问题： 20 | 21 | 最大化$$g(\lambda,v)$$ 22 | 23 | 不等式条件：$$\lambda_i>=0$$![](/assets/touyuhua1.png)![](/assets/touyouhua2.png)![](/assets/touyouhua3.png)![](/assets/touyouhua4.png)![](/assets/touyouhua5.png)![](/assets/touyouhua6.png) 24 | 25 | ![](/assets/touyouhua7.png)![](/assets/touyouhua8.png)![](/assets/touyouhua9.png)![](/assets/touyouhua10.png)![](/assets/touyouhua11.png) 26 | 27 | -------------------------------------------------------------------------------- /dl/reinforcement/banditsuan-fa.md: -------------------------------------------------------------------------------- 1 | # Thompson sampling算法 2 | 3 | 假设每个臂是否产生收益，其背后有一个概率分布，产生收益的概率为p 4 | 5 | 我们不断地试验，去估计出一个置信度较高的\*概率p的概率分布\*就能近似解决这个问题了。 6 | 7 | 怎么能估计概率p的概率分布呢？ 答案是假设概率p的概率分布符合beta\(wins, lose\)分布，它有两个参数: wins, lose。 8 | 9 | 每个臂都维护一个beta分布的参数。每次试验后，选中一个臂，摇一下，有收益则该臂的wins增加1，否则该臂的lose增加1。 10 | 11 | 每次选择臂的方式是：用每个臂现有的beta分布产生一个随机数b，选择所有臂产生的随机数中最大的那个臂去摇。 12 | 13 | 以上就是Thompson采样，用python实现就一行： 14 | 15 | ``` 16 | choice = numpy.argmax(pymc.rbeta(1 + self.wins, 1 + self.trials - self.wins)) 17 | ``` 18 | 19 | # Upper Confidence Bound\(置信区间上界\) 20 | 21 | 先对每一个臂都试一遍 22 | 23 | 之后，每次选择以下值最大的那个臂 24 | 25 | ![](/assets/ubc1.png) 26 | 27 | 其中加号前面是这个臂到目前的收益均值，后面的叫做bonus，本质上是均值的标准差，t是目前的试验次数，Tjt是这个臂被试次数。 28 | 29 | 这个公式反映：均值越大，标准差越小，被选中的概率会越来越大，起到了exploit的作用；同时哪些被选次数较少的臂也会得到试验机会，起到了explore的作用。 30 | 31 | # Epsilon-Greedy算法 32 | 33 | 选一个\(0,1\)之间较小的数epsilon 34 | 35 | 每次以概率epsilon（产生一个\[0,1\]之间的随机数，比epsilon小）做一件事：所有臂中随机选一个。否则，选择截止当前，平均收益最大的那个臂。 36 | 37 | 是不是简单粗暴？epsilon的值可以控制对Exploit和Explore的偏好程度。越接近0，越保守，只想花钱不想挣钱。 38 | 39 | -------------------------------------------------------------------------------- /dl/reinforcement/actor-critic.md: -------------------------------------------------------------------------------- 1 | Actor 和 Critic, 他们都能用不同的神经网络来代替 . 在 Policy Gradients 的影片中提到过, 现实中的奖惩会左右 Actor 的更新情况. Policy Gradients 也是靠着这个来获取适宜的更新. 那么何时会有奖惩这种信息能不能被学习呢? 这看起来不就是 以值为基础的强化学习方法做过的事吗. 那我们就拿一个 Critic 去学习这些奖惩机制, 学习完了以后. 由 Actor 来指手画脚, 由 Critic 来告诉 Actor 你的那些指手画脚哪些指得好, 哪些指得差, Critic 通过学习环境和奖励之间的关系, 能看到现在所处状态的潜在奖励, 所以用它来指点 Actor 便能使 Actor 每一步都在更新, 如果使用单纯的 Policy Gradients, Actor 只能等到回合结束才能开始更新. 2 | 3 | 4 | 5 | **一句话概括 Actor Critic 方法**: 6 | 7 | 结合了 Policy Gradient \(Actor\) 和 Function Approximation \(Critic\) 的方法.`Actor`基于概率选行为,`Critic`基于`Actor`的行为评判行为的得分,`Actor`根据`Critic`的评分修改选行为的概率. 8 | 9 | **Actor Critic 方法的优势**: 可以进行单步更新, 比传统的 Policy Gradient 要快. 10 | 11 | **Actor Critic 方法的劣势**: 取决于 Critic 的价值判断, 但是 Critic 难收敛, 再加上 Actor 的更新, 就更难收敛. 为了解决收敛问题, Google Deepmind 提出了`Actor Critic`升级版`Deep Deterministic Policy Gradient`. 后者融合了 DQN 的优势, 解决了收敛难的问题. 我们之后也会要讲到[Deep Deterministic Policy Gradient](https://morvanzhou.github.io/tutorials/machine-learning/reinforcement-learning/6-2-DDPG/). 不过那个是要以`Actor Critic`为基础, 懂了`Actor Critic`, 后面那个就好懂了. 12 | 13 | 14 | 15 | -------------------------------------------------------------------------------- /dl/layers/userdefine.md: -------------------------------------------------------------------------------- 1 | 对于简单的定制操作，我们或许可以通过使用 2 | 3 | `layers.core.Lambda` 4 | 5 | 层来完成。但对于任何具有可训练权重的定制层，你应该自己来实现。 6 | 7 | ``` 8 | from keras import backend as K 9 | from keras.engine.topology import Layer 10 | import numpy as np 11 | 12 | class MyLayer(Layer): 13 | 14 | def __init__(self, output_dim, **kwargs): 15 | self.output_dim = output_dim 16 | super(MyLayer, self).__init__(**kwargs) 17 | 18 | def build(self, input_shape): 19 | # Create a trainable weight variable for this layer. 20 | self.kernel = self.add_weight(name='kernel', 21 | shape=(input_shape[1], self.output_dim), 22 | initializer='uniform', 23 | trainable=True) 24 | super(MyLayer, self).build(input_shape) # Be sure to call this somewhere! 25 | 26 | def call(self, x): 27 | return K.dot(x, self.kernel) 28 | 29 | def compute_output_shape(self, input_shape): 30 | return (input_shape[0], self.output_dim) 31 | ``` 32 | 33 | 34 | 35 | -------------------------------------------------------------------------------- /dl/layers/rnn.md: -------------------------------------------------------------------------------- 1 | ## Recurrent层 {#recurrent_1} 2 | 3 | --- 4 | 5 | 这是循环层的抽象类，请不要在模型中直接应用该层（因为它是抽象类，无法实例化任何对象）。请使用它的子类`LSTM`，`GRU`或`SimpleRNN`。 6 | 7 | 所有的循环层（`LSTM`,`GRU`,`SimpleRNN`）都继承本层，因此下面的参数可以在任何循环层中使用 8 | 9 | ## SimpleRNN层 {#simplernn} 10 | 11 | --- 12 | 13 | 全连接RNN网络，RNN的输出会被回馈到输入 14 | 15 | ## GRU层 {#gru} 16 | 17 | --- 18 | 19 | 门限循环单元 20 | 21 | ## LSTM层 {#lstm} 22 | 23 | --- 24 | 25 | Keras长短期记忆模型 26 | 27 | ## ConvLSTM2D层 {#convlstm2d} 28 | 29 | --- 30 | 31 | ConvLSTM2D是一个LSTM网络，但它的输入变换和循环变换是通过卷积实现的 32 | 33 | ## SimpleRNNCell层 {#simplernncell} 34 | 35 | --- 36 | 37 | SinpleRNN的Cell类 38 | 39 | ## GRUCell层 {#grucell} 40 | 41 | --- 42 | 43 | GRU的Cell类 44 | 45 | ## LSTMCell层 {#lstmcell} 46 | 47 | --- 48 | 49 | LSTM的Cell类 50 | 51 | ## StackedRNNCells层 {#stackedrnncells} 52 | 53 | --- 54 | 55 | 这是一个wrapper，用于将多个recurrent cell包装起来，使其行为类型单个cell。该层用于实现搞笑的stacked RNN 56 | 57 | ## CuDNNGRU层 {#cudnngru} 58 | 59 | --- 60 | 61 | 基于CuDNN的快速GRU实现，只能在GPU上运行，只能使用tensoflow为后端 62 | 63 | ## CuDNNLSTM层 {#cudnnlstm} 64 | 65 | --- 66 | 67 | 基于CuDNN的快速LSTM实现，只能在GPU上运行，只能使用tensoflow为后端 68 | 69 | 70 | 71 | -------------------------------------------------------------------------------- /dl/reinforcement/ddpg.md: -------------------------------------------------------------------------------- 1 | Google DeepMind 提出的一种使用`Actor Critic`结构, 但是输出的不是行为的概率, 而是具体的行为, 用于连续动作 \(continuous action\) 的预测.`DDPG`结合了之前获得成功的`DQN`结构, 提高了`Actor Critic`的稳定性和收敛性. 2 | 3 | 4 | 5 | ## 算法 {#算法} 6 | 7 | `DDPG`的算法实际上就是一种`Actor Critic`, 8 | 9 | [![](https://morvanzhou.github.io/static/results/reinforcement-learning/6-2-0.png "Deep Deterministic Policy Gradient \(DDPG\) \(Tensorflow\)-0")](https://morvanzhou.github.io/static/results/reinforcement-learning/6-2-0.png) 10 | 11 | 关于`Actor`部分, 他的参数更新同样会涉及到`Critic`, 上面是关于`Actor`参数的更新, 它的前半部分`grad[Q]`是从`Critic`来的, 这是在说:**这次`Actor`的动作要怎么移动, 才能获得更大的`Q`**, 而后半部分`grad[u]`是从`Actor`来的, 这是在说:**`Actor`要怎么样修改自身参数, 使得`Actor`更有可能做这个动作**. 所以两者合起来就是在说:**`Actor`要朝着更有可能获取大`Q`的方向修改动作参数了**. 12 | 13 | [![](https://morvanzhou.github.io/static/results/reinforcement-learning/6-2-1.png "Deep Deterministic Policy Gradient \(DDPG\) \(Tensorflow\)-1")](https://morvanzhou.github.io/static/results/reinforcement-learning/6-2-1.png) 14 | 15 | 上面这个是关于`Critic`的更新, 它借鉴了`DQN`和`Double Q learning`的方式, 有两个计算`Q`的神经网络,`Q_target`中依据下一状态, 用`Actor`来选择动作, 而这时的`Actor`也是一个`Actor_target`\(有着 Actor 很久之前的参数\). 使用这种方法获得的`Q_target`能像`DQN`那样切断相关性, 提高收敛性. 16 | 17 | -------------------------------------------------------------------------------- /ml/cluster/gmm.md: -------------------------------------------------------------------------------- 1 | 1）概述 2 | 3 | 正态分布也叫高斯分布，正态分布的概率密度曲线也叫高斯分布概率曲线_。_ 4 | 5 | GaussianMixtureModel\(混合高斯模型，GMM\)。 6 | 7 | 聚类算法大多数通过相似度来判断，而相似度又大多采用欧式距离长短作为衡量依据。而GMM采用了新的判断依据：概率，即通过属于某一类的概率大小来判断最终的归属类别。 8 | 9 | GMM的基本思想就是：任意形状的概率分布都可以用多个高斯分布函数去近似，也就是说GMM就是有多个单高斯密度分布（Gaussian）组成的，每个Gaussian叫一个"Component"，这些"Component"线性加成在一起就组成了 GMM 的概率密度函数，也就是下面的函数。 10 | 11 | 2）数学公式 12 | 13 | ![](http://images2015.cnblogs.com/blog/1119747/201706/1119747-20170612143601915-492097161.png) 14 | 15 | K：模型的个数，即Component的个数（聚类的个数） 16 | 17 | ![](http://images2015.cnblogs.com/blog/1119747/201706/1119747-20170612145911306-173735240.png)为第k个高斯的权重 18 | 19 | p（x \|k） 则为第k个高斯概率密度,其均值为μk,方差为σk 20 | 21 | 上述参数，除了K是直接给定之外，其他参数都是通过EM算法估算出来的。\(有个参数是指定EM算法参数的\) 22 | 23 | 3）GaussianMixtureModel 算法函数 24 | 25 | a）from sklearn.mixture.GaussianMixture 26 | 27 | b）主要参数（[详细参数](http://scikit-learn.org/dev/modules/generated/sklearn.mixture.GaussianMixture.html#sklearn.mixture.GaussianMixture)） 28 | 29 | n\_components ：高斯模型的个数，即聚类的目标个数 30 | 31 | covariance\_type : 通过EM算法估算参数时使用的协方差类型，默认是"full" 32 | 33 | full：每个模型使用自己的一般协方差矩阵 34 | 35 | tied：所用模型共享一个一般协方差矩阵 36 | 37 | diag：每个模型使用自己的对角线协方差矩阵 38 | 39 | spherical：每个模型使用自己的单一方差 40 | 41 | -------------------------------------------------------------------------------- /mian-shi-ti-mu-1.md: -------------------------------------------------------------------------------- 1 | **问题1：** 2 | 3 | **阐述批归一化的意义** 4 | 5 | 这是一个非常好的问题，因为这涵盖了面试者在操作神经网络模型时所需知道的大部分知识。你的回答方式可以不同，但都需要说明以下主要思想： 6 | 7 | 8 | 9 | 10 | ![](https://mmbiz.qpic.cn/mmbiz_png/vI9nYe94fsFIyfpbd6KHUrNNBQ59hcRW12bLywuLAibUBzoqXDJrAhsuVia0U6S1piar3H6xiabXiakEOLZiah0Qonicg/640?wx_fmt=png&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1) 11 | 12 | _算法 1：_ 13 | 14 | _批归一化变换，在一个 mini-batch 上应用于激活 x。_ 15 | 16 | 批归一化是一种用于训练神经网络模型的有效方法。这种方法的目标是对特征进行归一化处理（使每层网络的输出都经过激活），得到标准差为 1 的零均值状态。所以其相反的现象是非零均值。这将如何影响模型的训练： 17 | 18 | 首先，这可以被理解成非零均值是数据不围绕 0 值分布的现象，而是数据的大多数值大于 0 或小于 0。结合高方差问题，数据会变得非常大或非常小。在训练层数很多的神经网络时，这个问题很常见。如果特征不是分布在稳定的区间（从小到大的值）里，那么就会对网络的优化过程产生影响。我们都知道，优化神经网络将需要用到导数计算。 19 | 20 | 21 | 22 | 23 | 假设一个简单的层计算公式 y = \(Wx + b\)，y 在 W 上的导数就是这样：dy=dWx。因此，x 的值会直接影响导数的值（当然，神经网络模型的梯度概念不会如此之简单，但理论上，x 会影响导数）。因此，如果 x 引入了不稳定的变化，则这个导数要么过大，要么就过小，最终导致学习到的模型不稳定。而这也意味着当使用批归一化时，我们可以在训练中使用更高的学习率。 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | 批归一化可帮助我们避免 x 的值在经过非线性激活函数之后陷入饱和的现象。也就是说，批归一化能够确保激活都不会过高或过低。这有助于权重学习——如果不使用这一方案，某些权重可能永远不会学习。这还能帮助我们降低对参数的初始值的依赖。 32 | 33 | 34 | 35 | 36 | 批归一化也可用作正则化（regularization）的一种形式，有助于实现过拟合的最小化。使用批归一化时，我们无需再使用过多的 dropout；这是很有助益的，因为我们无需担心再执行 dropout 时丢失太多信息。但是，仍然建议组合使用这两种技术。 37 | 38 | -------------------------------------------------------------------------------- /dl/introduction/chapter1.md: -------------------------------------------------------------------------------- 1 | # 第一章 使用神经网络识别手写数字 2 | 3 | 人类视觉系统是世界上众多奇迹之一。看看下面的手写数字序列： 4 | 5 | ![](/images/1.png) 6 | 7 | 大多数人毫不费力就能够认出这些数字为 504192. 这么容易反而让人觉着迷惑了。在人类的每个脑半球中，有着一个初级视觉皮层，常称为 V1，包含 1 亿 4 千万个神经元及数百亿条神经元间的连接。但是人类视觉不是就只有 V1，还包括整个视觉皮层——V2、V3、V4 和 V5——他们逐步地进行更加复杂的图像处理。人类的头脑就是一台超级计算机，通过数十亿年的进化不断地演变，最终能够极好地适应理解视觉世界的任务。识别手写数字也不是一件简单的事。尽管人类在理解我们眼睛展示出来的信息上非常擅长，但几乎所有的过程都是无意识地。所以，我们通常并不能体会自身视觉系统解决问题的困难。 8 | 9 | 如果你尝试写出计算机程序来识别诸如上面的数字，就会明显感受到视觉模式识别的困难。看起来人类一下子就能完成的任务变得特别困难。关于我们识别形状——“9 顶上有一个圈，右下方则是一条竖线”这样的简单直觉——实际上算法上就很难轻易表达出来了。而在你试着让这些识别规则越发精准时，就会很快陷入各种混乱的异常或者特殊情形的困境中。看起来毫无希望。 10 | 11 | 神经网络以另一种方式看待这个问题。其主要思想是获取大量的手写数字，常称作*训练样本*， 12 | 13 | ![](/images/2.png) 14 | 15 | 然后开发出一个可以从这些训练样本中进行学习的系统。换言之，神经网络使用样本来自动推断出识别手写数字的规则。另外，通过增加训练样本的数量，网络可以学到更多关于手写数字的知识，这样就能够提升自身的准确性。所以，上面例子中我们只是展出了 100 个训练数字样本，而通过使用数千或者数百万或者数十亿的训练样本我们也许能够得到更好的手写数字识别器。 16 | 17 | 本章我们将实现一个可以识别手写数字的神经网络。这个程序仅仅 74 行，不适用特别的神经网络库。然而，这个短小的网络不需要人类帮助便可以超过 96% 的准确率识别数字。而且，在后面的章节，我们会发展出将准确率提升到 99% 的技术。实际上，最优的商业神经网络已经足够好到被银行和邮局分别用在账单核查和识别地址上了。 18 | 19 | 手写识别常常被当成学习神经网络的原型问题，因此我们聚焦在这个问题上。作为一个原型，它具备一个关键点：挑战性——识别手写数字并不轻松——但也不会难到需要超级复杂的解决方法，或者超大规模的计算资源。另外，这其实也是一种发展出诸如深度学习更加高级的技术的方法。所以，整本书我们都会持续地讨论手写数字识别问题。本书后面部分，我们会讨论这些想法如何用在其他计算机视觉的问题或者语音、自然语言处理和其他一些领域中。 20 | 21 | [待续] -------------------------------------------------------------------------------- /dl/layers/regular.md: -------------------------------------------------------------------------------- 1 | Batch normalization 的 batch 是批数据, 把数据分成小批小批进行 stochastic gradient descent. 而且在每批数据进行前向传递 forward propagation 的时候, 对每一层都进行 normalization 的处理, 2 | 3 | Batch normalization 也可以被看做一个层面. 在一层层的添加神经网络的时候, 我们先有数据 X, 再添加全连接层, 全连接层的计算结果会经过 激励函数 成为下一层的输入, 接着重复之前的操作. Batch Normalization \(BN\) 就被添加在每一个全连接和激励函数之间 4 | 5 | ![](/assets/deeplayer-bn1.png) 6 | 7 | ![](/assets/deeplayer-bn2.png) 8 | 9 | 没有 normalize 的数据 使用 tanh 激活以后, 激活值大部分都分布到了饱和阶段, 也就是大部分的激活值不是-1, 就是1, 而 normalize 以后, 大部分的激活值在每个分布区间都还有存在. 再将这个激活后的分布传递到下一层神经网络进行后续计算, 每个区间都有分布的这一种对于神经网络就会更加有价值. Batch normalization 不仅仅 normalize 了一下数据, 他还进行了反 normalize 的手续. 为什么要这样呢? 10 | 11 | ## ![](/assets/deeplayer-bn3.png) {#batchnormalization_1} 12 | 13 | 我们引入一些 batch normalization 的公式. 这三步就是我们在刚刚一直说的 normalization 工序, 但是公式的后面还有一个反向操作, 将 normalize 后的数据再扩展和平移. 原来这是为了让神经网络自己去学着使用和修改这个扩展参数 gamma, 和 平移参数 β, 这样神经网络就能自己慢慢琢磨出前面的 normalization 操作到底有没有起到优化的作用, 如果没有起到作用, 我就使用 gamma 和 belt 来抵消一些 normalization 的操作. 14 | 15 | ![](/assets/5_13_03.gif) 16 | 17 | 可以看出, 没有用 BN 的时候, 每层的值迅速全部都变为 0, 也可以说, 所有的神经元都已经死了. 而有 BN,`relu`过后, 每层的值都能有一个比较好的分布效果, 大部分神经元都还活着. 18 | 19 | 20 | 21 | ![](/assets/5_13_06.gif) 22 | 23 | 可以看出, 没有 NB, 每层的值迅速全部都饱和, 都跑去了 -1/1 这个饱和区间, 有 NB, 即使前一层因变得相对饱和, 但是后面几层的值都被 normalize 到有效的不饱和区间内计算. 确保了一个活的神经网络. 24 | 25 | ## BatchNormalization层 {#batchnormalization_1} 26 | 27 | --- 28 | 29 | 该层在每个batch上将前一层的激活值重新规范化，即使得其输出数据的均值接近0，其标准差接近1 30 | 31 | -------------------------------------------------------------------------------- /math/analytic/cross-validation.md: -------------------------------------------------------------------------------- 1 | ## 交叉验证 2 | 3 | --- 4 | 5 | 交叉验证是在机器学习建立模型和验证模型参数时常用的办法。交叉验证，顾名思义，就是重复的使用数据，把得到的样本数据进行切分，组合为不同的训练集和测试集，用训练集来训练模型，用测试集来评估模型预测的好坏。在此基础上可以得到多组不同的训练集和测试集，某次训练集中的某样本在下次可能成为测试集中的样本，即所谓“交叉”。 6 | 7 | 那么什么时候才需要交叉验证呢？交叉验证用在数据不是很充足的时候。比如在我日常项目里面，对于普通适中问题，如果数据样本量小于一万条，我们就会采用交叉验证来训练优化选择模型。如果样本大于一万条的话，我们一般随机的把数据分成三份，一份为训练集（Training Set），一份为验证集（Validation Set），最后一份为测试集（Test Set）。用训练集来训练模型，用验证集来评估模型预测的好坏和选择模型及其对应的参数。把最终得到的模型再用于测试集，最终决定使用哪个模型以及对应参数。 8 | 9 | 回到交叉验证，根据切分的方法不同，交叉验证分为下面三种： 10 | 11 | 第一种是**简单交叉验证**，所谓的简单，是和其他交叉验证方法相对而言的。首先，我们随机的将样本数据分为两部分（比如： 70%的训练集，30%的测试集），然后用训练集来训练模型，在测试集上验证模型及参数。接着，我们再把样本打乱，重新选择训练集和测试集，继续训练数据和检验模型。最后我们选择损失函数评估最优的模型和参数。 12 | 13 | 第二种是**S折交叉验证**（S-Folder Cross Validation）。和第一种方法不同，S折交叉验证会把样本数据随机的分成S份，每次随机的选择S-1份作为训练集，剩下的1份做测试集。当这一轮完成后，重新随机选择S-1份来训练数据。若干轮（小于S）之后，选择损失函数评估最优的模型和参数。 14 | 15 | 第三种是**留一交叉验证**（Leave-one-out Cross Validation），它是第二种情况的特例，此时S等于样本数N，这样对于N个样本，每次选择N-1个样本来训练数据，留一个样本来验证模型预测的好坏。此方法主要用于样本量非常少的情况，比如对于普通适中问题，N小于50时，我一般采用留一交叉验证。 16 | 17 | 通过反复的交叉验证，用损失函数来度量得到的模型的好坏，最终我们可以得到一个较好的模型。那这三种情况，到底我们应该选择哪一种方法呢？一句话总结，如果我们只是对数据做一个初步的模型建立，不是要做深入分析的话，简单交叉验证就可以了。否则就用S折交叉验证。在样本量少的时候，使用S折交叉验证的特例留一交叉验证。 18 | 19 | 此外还有一种比较特殊的交叉验证方式，也是用于样本量少的时候。叫做**自助法**\(bootstrapping\)。比如我们有m个样本（m较小），每次在这m个样本中随机采集一个样本，放入训练集，采样完后把样本放回。这样重复采集m次，我们得到m个样本组成的训练集。当然，这m个样本中很有可能有重复的样本数据。同时，用原始的m个样本做测试集。这样接着进行交叉验证。由于我们的训练集有重复数据，这会改变数据的分布，因而训练结果会有估计偏差，因此，此种方法不是很常用，除非数据量真的很少，比如小于20个。 20 | 21 | -------------------------------------------------------------------------------- /dl/layers/ebbedded.md: -------------------------------------------------------------------------------- 1 | ## Embedding层 {#embedding_1} 2 | 3 | --- 4 | 5 | 嵌入层将正整数（下标）转换为具有固定大小的向量，如\[\[4\],\[20\]\]->\[\[0.25,0.1\],\[0.6,-0.2\]\] 6 | 7 | Embedding层只能作为模型的第一层。 8 | 9 | 较为费劲的就是第一句话： 10 | _**嵌入层将正整数（下标）转换为具有固定大小的向量，如\[\[4\],\[20\]\]->\[\[0.25,0.1\],\[0.6,-0.2\]\]**_ 11 | 12 | 哪到底咋转啊，亲？ 13 | 14 | 这涉及到词向量，具体看可以参考[Word2vec](/dl/word2vec/word2vec.md) 15 | 16 | ![](/assets/deeplayerr-embemding1.png) 17 | 18 | 上图的流程是把文章的单词使用词向量来表示。 19 | \(1\)提取文章所有的单词，把其按其出现的次数降序\(这里只取前50000个\)，比如单词‘network’出现的次数最多，编号ID为0，依次类推… 20 | 21 | \(2\)每个编号ID都可以使用50000维的二进制\(one-hot\)表示 22 | 23 | \(3\)最后，我们会生产一个矩阵M，行大小为词的个数50000，列大小为词向量的维度\(通常取128或300\)，比如矩阵的第一行就是编号ID=0，即network对应的词向量。 24 | 25 | 那这个矩阵M怎么获得呢？在Skip-Gram 模型中，我们会随机初始化它，然后使用神经网络来训练这个权重矩阵 26 | 27 | ![](/assets/deeplayer-embeding2.png) 28 | 29 | 那我们的输入数据和标签是什么？如下图，输入数据就是中间的哪个蓝色的词对应的one-hot编码，标签就是它附近词的one-hot编码\(这里windown\_size=2,左右各取2个\) 30 | 31 | ![](/assets/deeplayer-embeding3.png) 32 | 33 | 就上述的Word2Vec中的demo而言，它的单词表大小为1000，词向量的维度为300，所以Embedding的参数 input\_dim=10000，output\_dim=300 34 | 35 | 回到最初的问题：_**嵌入层将正整数（下标）转换为具有固定大小的向量，如\[\[4\],\[20\]\]->\[\[0.25,0.1\],\[0.6,-0.2\]\]**_ 36 | 37 | 举个栗子：假如单词表的大小为1000，词向量维度为2，经单词频数统计后，tom对应的id=4，而jerry对应的id=20，经上述的转换后，我们会得到一个M1000×2的矩阵，而tom对应的是该矩阵的第4行，取出该行的数据就是\[0.25,0.1\] 38 | 39 | 如果输入数据不需要词的语义特征语义，简单使用Embedding层就可以得到一个对应的词向量矩阵，但如果需要语义特征，我们大可把以前训练好的词向量权重直接扔到Embedding层中即可，具体看参考keras提供的栗子:[在Keras模型中使用预训练的词向量](https://github.com/MoyanZitto/keras-cn/blob/master/docs/legacy/blog/word_embedding.md) 40 | 41 | -------------------------------------------------------------------------------- /dl/reinforcement/policy-gridient.md: -------------------------------------------------------------------------------- 1 | 强化学习是一个通过奖惩来学习正确行为的机制. 家族中有很多种不一样的成员, 有学习奖惩值, 根据自己认为的高价值选行为, 比如[Q learning](https://morvanzhou.github.io/tutorials/machine-learning/ML-intro/4-03-q-learning/)[Deep Q Network](https://morvanzhou.github.io/tutorials/machine-learning/ML-intro/4-06-DQN/), 也有不通过分析奖励值, 直接输出行为的方法, 这就是今天要说的 Policy Gradients 了. 甚至我们可以为 Policy Gradients 加上一个神经网络来输出预测的动作. 对比起以值为基础的方法, Policy Gradients 直接输出动作的最大好处就是, 它能在一个连续区间内挑选动作, 而基于值的, 比如 Q-learning, 它如果在无穷多的动作中计算价值, 从而选择行为, 这, 它可吃不消. 2 | 3 | 有了神经网络当然方便, 但是, 我们怎么进行神经网络的误差反向传递呢? Policy Gradients 的误差又是什么呢? 答案是! 哈哈, 没有误差! 但是他的确是在进行某一种的反向传递. 这种反向传递的目的是让这次被选中的行为更有可能在下次发生. 但是我们要怎么确定这个行为是不是应当被增加被选的概率呢? 这时候我们的老朋友, reward 奖惩正可以在这时候派上用场, 4 | 5 | 现在我们来演示一遍, 观测的信息通过神经网络分析, 选出了左边的行为, 我们直接进行反向传递, 使之下次被选的可能性增加, 但是奖惩信息却告诉我们, 这次的行为是不好的, 那我们的动作可能性增加的幅度 随之被减低. 这样就能靠奖励来左右我们的神经网络反向传递. 我们再来举个例子, 假如这次的观测信息让神经网络选择了右边的行为, 右边的行为随之想要进行反向传递, 使右边的行为下次被多选一点, 这时, 奖惩信息也来了, 告诉我们这是好行为, 那我们就在这次反向传递的时候加大力度, 让它下次被多选的幅度更猛烈! 这就是 Policy Gradients 的核心思想了. 很简单吧. 6 | 7 | Policy gradient 是 RL 中另外一个大家族, 他不像 Value-based 方法 \(Q learning, Sarsa\), 但他也要接受环境信息 \(observation\), 不同的是他要输出不是 action 的 value, 而是具体的那一个 action, 这样 policy gradient 就跳过了 value 这个阶段. 而且个人认为 Policy gradient 最大的一个优势是: 输出的这个 action 可以是一个连续的值, 之前我们说到的 value-based 方法输出的都是不连续的值, 然后再选择值最大的 action. 而 policy gradient 可以在一个连续分布上选取 action. 8 | 9 | ## 算法 {#算法} 10 | 11 | 我们介绍的 policy gradient 的第一个算法是一种基于**整条回合数据**的更新, 也叫**REINFORCE**方法. 这种方法是 policy gradient 的最基本方法, 有了这个的基础, 我们再来做更高级的. 12 | 13 | ![](/assets/reinforcement-pg1.png) 14 | 15 | 16 | 17 | -------------------------------------------------------------------------------- /ml/recommand/mr-itemcf.md: -------------------------------------------------------------------------------- 1 | ``` 2 | Sample input 3 | ``` 4 | 5 | ``` 6 | +----+---+---+ 7 | | u0 | A | 2 | 8 | +----+---+---+ 9 | | u0 | C | 1 | 10 | +----+---+---+ 11 | | u1 | A | 7 | 12 | +----+---+---+ 13 | | u1 | B | 1 | 14 | +----+---+---+ 15 | | u2 | A | 2 | 16 | +----+---+---+ 17 | | u2 | C | 1 | 18 | +----+---+---+ 19 | | u0 | B | 2 | 20 | +----+---+---+ 21 | Format is (user, item, pref). 22 | 23 | MapReduce1 24 | (user, item, pref) = 25 | > 26 | (user, [(item, pref)]) 27 | 28 | (u0, [(A, 2), (B, 2), (C, 1)]) 29 | (u1, [(A, 7), (B, 1)]) 30 | (u2, [(A, 2), (C, 1)]) 31 | 32 | 33 | MapReduce2: 34 | Map: 35 | (user, [(item, pref)]) = 36 | > 37 | {(item1, item2): (pref1, pref2)} 38 | 39 | {(A, B): (2, 2)} 40 | {(A, C): (2, 1)} 41 | {(B, C): (2, 1)} 42 | ---------------- 43 | {(A, B): (7, 1)} 44 | ---------------- 45 | {(A, C): (2, 1)} 46 | 47 | Group: 48 | 49 | Reduce: 50 | {(item1, item2): [(pref1, pref2)]} = 51 | > 52 | 53 | {(item1, sim): (item1, item2, sim)} 54 | {(item2, sim): (item1, item2, sim)} 55 | 56 | Input: 57 | {(A, B): [(2, 2), (7, 1)]} 58 | ---------------- 59 | {(A, C): [(2, 1), (2, 1)]} 60 | ---------------- 61 | {(B, C): [(2, 1)]} 62 | 63 | Output: 64 | The sim is faked. For sample size 2, Pearson Coefficient is always 1 or -1. 65 | For data is needed to illustrate similarities. 66 | 67 | {(A, 0.5): (A, B, 0.5)} 68 | {(B, 0.5): (B, A, 0.5)} 69 | {(A, 0.4): (A, C, 0.4)} 70 | {(C, 0.4): (C, A, 0.4)} 71 | {(B, 0.3): (B, C, 0.3)} 72 | {(C, 0.3): (C, B, 0.3)} 73 | 74 | MapReduce3: 75 | group on item, sort on (item, sim) 76 | ``` 77 | 78 | 79 | 80 | -------------------------------------------------------------------------------- /ml/cluster/knn-code.md: -------------------------------------------------------------------------------- 1 | ```py 2 | #encoding:utf-8 3 | from numpy import * 4 | import operator 5 | 6 | def createDataSet(): 7 | group = array([[1.0,1.1],[1.0,1.0],[0,0],[0,0.1]]) 8 | labels = ['A','A','B','B'] 9 | return group,labels 10 | 11 | def classify(inX,dataSet,labels,k): 12 | #返回“数组”的行数，如果shape[1]返回的则是数组的列数 13 | dataSetSize = dataSet.shape[0] 14 | #两个“数组”相减，得到新的数组 15 | diffMat = tile(inX,(dataSetSize,1))- dataSet 16 | #求平方 17 | sqDiffMat = diffMat **2 18 | #求和，返回的是一维数组 19 | sqDistances = sqDiffMat.sum(axis=1) 20 | #开方，即测试点到其余各个点的距离 21 | distances = sqDistances **0.5 22 | #排序，返回值是原数组从小到大排序的下标值 23 | sortedDistIndicies = distances.argsort() 24 | #定义一个空的字典 25 | classCount = {} 26 | for i in range(k): 27 | #返回距离最近的k个点所对应的标签值 28 | voteIlabel = labels[sortedDistIndicies[i]] 29 | #存放到字典中 30 | classCount[voteIlabel] = classCount.get(voteIlabel,0)+1 31 | #排序 classCount.iteritems() 输出键值对 key代表排序的关键字 True代表降序 32 | sortedClassCount = sorted(classCount.iteritems(),key = operator.itemgetter(1),reverse = True) 33 | #返回距离最小的点对应的标签 34 | return sortedClassCount[0][0] 35 | ``` 36 | 37 | ``` 38 | import kNN 39 | from numpy import * 40 | 41 | dataSet, labels = kNN.createDataSet() 42 | 43 | testX = array([1.2, 1.0]) 44 | k = 3 45 | outputLabel = kNN.kNNClassify(testX, dataSet, labels, 3) 46 | print "Your input is:", testX, "and classified to class: ", outputLabel 47 | 48 | testX = array([0.1, 0.3]) 49 | outputLabel = kNN.kNNClassify(testX, dataSet, labels, 3) 50 | print "Your input is:", testX, "and classified to class: ", outputLabel 51 | ``` 52 | 53 | 54 | 55 | -------------------------------------------------------------------------------- /ml/bayes/simple-bayes-real-use.md: -------------------------------------------------------------------------------- 1 | ## 朴素贝叶斯分类实例：检测SNS社区中不真实账号 2 | 3 | 下面讨论一个使用朴素贝叶斯分类解决实际问题的例子，为了简单起见，对例子中的数据做了适当的简化。 4 | 5 | 这个问题是这样的，对于SNS社区来说，不真实账号（使用虚假身份或用户的小号）是一个普遍存在的问题，作为SNS社区的运营商，希望可以检测出这些不真实账号，从而在一些运营分析报告中避免这些账号的干扰，亦可以加强对SNS社区的了解与监管。 6 | 7 | 如果通过纯人工检测，需要耗费大量的人力，效率也十分低下，如能引入自动检测机制，必将大大提升工作效率。这个问题说白了，就是要将社区中所有账号在真实账号和不真实账号两个类别上进行分类，下面我们一步一步实现这个过程。 8 | 9 | 首先设C=0表示真实账号，C=1表示不真实账号。 10 | 11 | _1、确定特征属性及划分_ 12 | 13 | 这一步要找出可以帮助我们区分真实账号与不真实账号的特征属性，在实际应用中，特征属性的数量是很多的，划分也会比较细致，但这里为了简单起见，我们用少量的特征属性以及较粗的划分，并对数据做了修改。 14 | 15 | 我们选择三个特征属性：a1：日志数量/注册天数，a2：好友数量/注册天数，a3：是否使用真实头像。在SNS社区中这三项都是可以直接从数据库里得到或计算出来的。 16 | 17 | 下面给出划分：a1：{a<=0.05, 0.05<a<0.2, a>=0.2}，a1：{a<=0.1, 0.1<a<0.8, a>=0.8}，a3：{a=0（不是）,a=1（是）}。 18 | 19 | _2、获取训练样本_ 20 | 21 | 这里使用运维人员曾经人工检测过的1万个账号作为训练样本。 22 | 23 | _3、计算训练样本中每个类别的频率_ 24 | 25 | 用训练样本中真实账号和不真实账号数量分别除以一万，得到： 26 | 27 | P\(C=0\)=8900/100000=0.89 28 | 29 | P\(C=1\)=110/100000=0.11 30 | 31 | _4、计算每个类别条件下各个特征属性划分的频率_ 32 | 33 | $$P(a_1<=0.05|C=0)=0.3$$ 34 | 35 | $$P(0.050.2|C=0)=0.2$$ 38 | 39 | $$P(a_1<0.05|C=1)=0.8$$ 40 | 41 | $$P(0.050.2|C=1)=0.1$$ 44 | 45 | $$P(a_2<=0.1|C=0)=0.1$$ 46 | 47 | $$P(0.10.8|C=0)=0.2$$ 50 | 51 | $$P(a_2<=0.1|C=1)=0.7$$ 52 | 53 | $$P(0.10.2|C=1)=0.1$$ 56 | 57 | $$P(a_3=0|C=0)=0.2$$ 58 | 59 | $$P(a_3=1|C=0)=0.8$$ 60 | 61 | $$P(a_3=0|C=1)=0.9$$ 62 | 63 | $$P(a_3=1|C=1)=0.1$$ 64 | 65 | _5、使用分类器进行鉴别_ 66 | 67 | 下面我们使用上面训练得到的分类器鉴别一个账号，这个账号使用非真实头像，日志数量与注册天数的比率为0.1，好友数与注册天数的比率为0.2。 68 | 69 | $$P(C=0)P(x|C=0)=P(C=0)P(0.05