Deep Learning ãªã³ã¯é
Deep Learningï¼æ·±å±¤å¦ç¿ï¼ã«é¢é£ããã¾ã¨ããã¼ã¸ã¨ãã¦ä½¿ç¨ããäºå®ã§ããDeep Learningã«é¢ããè¨äºã»ã¹ã©ã¤ãã»è«æã»åç»ã»æ¸ç±ã¸ã®ãªã³ã¯ãã¾ã¨ãã¦ãã¾ããææ°ã®ç 究ååã¯å ¨ç¶ææ¡ã§ãã¦ããªãã®ã§ä»å¾ç 究ãé²ãããªãã§è¨é²ãã¦ããããã¨æãã¾ããèªãã è«æã®æ¦è¦ãç°¡åã«ã¾ã¨ãã¦ããäºå®ã§ããæ¬ããã°ã§ã¯ãå½é¢ã®éãTheanoã使ã£ã¦å種Deep Learningã¢ã«ã´ãªãºã ãå®è£ ãã¦ããããã¨æãã¾ãã
é¢é£ãã¥ã¼ã¹ãªã©ã¯Twitterã§ãæµãã¦ããã®ã§èå³ããã£ãããã©ãã¼ãã¦ãã ããã
ãã¹ã¦ã«ç®ãéããæ´æ°ã追ãã¤ãã¦ãã¾ãããç§ã®ã¯ã¦ãªããã¯ãã¼ã¯ã§[Deep Learning]ã¨ããã¿ã°ãä»ãã¦ç»é²ãã¦ãã¾ããã¾ã£ããæ´çã§ãã¦ãã¾ããããåèã¾ã§ã
Theanoç·¨
- TheanoãWindowsã«ã¤ã³ã¹ãã¼ã«ï¼2015/1/27ï¼
- Theanoã®ä½¿ãæ¹ (1) ã·ã³ãã«ã¨å ±æå¤æ°ï¼2015/5/9ï¼
- Theanoã®ä½¿ãæ¹ (2) èªåå¾®åï¼2015/5/18ï¼
- Theanoã«ãã2ã¯ã©ã¹ãã¸ã¹ãã£ãã¯å帰ã®å®è£ ï¼2015/5/19ï¼
- Theanoã«ãããã¸ã¹ãã£ãã¯å帰ã®å®è£ ï¼2015/5/26ï¼
- Theanoã«ããå¤å±¤ãã¼ã»ãããã³ã®å®è£ ï¼2015/6/18ï¼
- Theanoã«ããç³ã¿è¾¼ã¿ãã¥ã¼ã©ã«ãããã¯ã¼ã¯ã®å®è£ (1)ï¼2015/6/26ï¼
- Theanoã«ããç³ã¿è¾¼ã¿ãã¥ã¼ã©ã«ãããã¯ã¼ã¯ã®å®è£ (2)ï¼2015/7/14ï¼
- Theanoã«ããèªå·±ç¬¦å·åå¨ï¼Autoencorderï¼ã®å®è£ ï¼2015/12/3ï¼
- Theanoã«ããéé³é¤å»èªå·±ç¬¦å·åå¨ï¼Denoising autoencoderï¼ã®å®è£ ï¼2015/12/9ï¼
- Theanoã«ããç©å±¤èªå·±ç¬¦å·åå¨ï¼Stacked autoencoderï¼ã®å®è£ ï¼2016/1/22ï¼
- ç©å±¤èªå·±ç¬¦å·åå¨ã®æ§è½è©ä¾¡ï¼2016/2/6ï¼
Chainerç·¨ï¼1.3.2ãã¼ã¹ãªã®ã§å¤ãã§ãï¼
- Chainerã«ããå¤å±¤ãã¼ã»ãããã³ã®å®è£ ï¼2015/10/5ï¼
- Chainerã«ããç³ã¿è¾¼ã¿ãã¥ã¼ã©ã«ãããã¯ã¼ã¯ã®å®è£ ï¼2015/10/7ï¼
- CIFAR-10ï¼2015/10/14ï¼
- Chainerã«ããCIFAR-10ã®ä¸è¬ç©ä½èªè (1)ï¼2015/11/8ï¼
- Chainerã«ããCIFAR-10ã®ä¸è¬ç©ä½èªè (2)ï¼2015/11/14ï¼
Kerasç·¨ NEW!
- 深層å¦ç¿ã©ã¤ãã©ãª Kerasï¼2016/3/28ï¼
- Kerasã¯ããã¾ãï¼2016/10/18ï¼
- Kerasã®ã¤ã³ã¹ãã¼ã«ï¼2016/10/24ï¼
- Kerasã«ãã2ã¯ã©ã¹ãã¸ã¹ãã£ãã¯å帰ï¼2016/10/30ï¼
- Kerasã«ãã2ã¯ã©ã¹åé¡ï¼Pima Indians Diabetesï¼ï¼2016/11/3ï¼
- Kerasã«ããå¤ã¯ã©ã¹åé¡ï¼Irisï¼ï¼2016/11/8ï¼
- Kerasã§MNISTï¼2016/11/9ï¼
- Kerasã«ããç³ã¿è¾¼ã¿ãã¥ã¼ã©ã«ãããã¯ã¼ã¯ã®å®è£ ï¼2016/11/20ï¼
- Kerasã§CIFAR-10ã®ä¸è¬ç©ä½èªèï¼2016/11/27ï¼
- Kerasã«ãããã¼ã¿æ¡å¼µï¼2016/12/12ï¼
- Kerasã§VGG16ã使ãï¼2017/1/4ï¼
- VGG16ã®Fine-tuningã«ããç¬ç«èªè (1)ï¼2017/1/8ï¼
- VGG16ã®Fine-tuningã«ããç¬ç«èªè (2)ï¼2017/1/10ï¼
- VGG16ã®Fine-tuningã«ãã17種é¡ã®è±ã®åé¡ï¼2017/1/31ï¼
- ç³ã¿è¾¼ã¿ãã¥ã¼ã©ã«ãããã¯ã¼ã¯ã®å¯è¦åï¼2017/2/16ï¼
ãã®ä»é¢é£è¨äº
- ãã¿ã¼ã³èªèã¨æ©æ¢°å¦ç¿ï¼PRMLï¼ã¾ã¨ãï¼2010/8/29ï¼
- Pythonã«ããã¢ã³ãã«ã«ãæ³å ¥éï¼2014/6/20ï¼
- 人工ç¥è½ãå®ç¾ããå¦ç¿ã¢ã«ã´ãªãºã ã«å¿ è¦ãªè½åï¼2014/9/10ï¼
- ãã«ããã³ãã·ã³ï¼å¯è¦å¤æ°ã®ã¿ï¼ã®å°åºï¼2016/3/11ï¼
- ãã«ããã³ãã·ã³ï¼é ãå¤æ°ããï¼ã®å°åºï¼2016/3/12ï¼
- å¶éãã«ããã³ãã·ã³ (RBM) ã®å°åº (1)ï¼2016/3/16ï¼
- å¶éãã«ããã³ãã·ã³ (RBM) ã®å°åº (2)ï¼2016/3/20ï¼
人工ç¥è½å¦ä¼ã®ç¹éè¨äº
ãµã¼ãã¤ã®ã¨ã£ãããã¨ãã¦æé©ã ããã©ãæ®å¿µãªããPDFãç¡æãã¦ã³ãã¼ãã§ããªãããã®ç¹éãå çã»åç·¨éããæ¸ç±ãçºå£²ããã¦ããï¼2015/11/5ï¼ã
- é£è¼è§£èª¬ãDeep Learningï¼æ·±å±¤å¦ç¿ï¼ãã«ããã£ã¦
- 第1å ãã£ã¼ããã«ããã³ãã·ã³å ¥é : ãã«ããã³ãã·ã³å¦ç¿ã®åºç¤
- 第2å å¤å±¤ãã¥ã¼ã©ã«ãããã¯ã¼ã¯ã«ãã深層表ç¾ã®å¦ç¿
- 第3å 大è¦æ¨¡Deep Learning(深層å¦ç¿)ã®å®ç¾æè¡
- 第4å ç»åèªèã®ããã®æ·±å±¤å¦ç¿
- 第5å é³å£°èªèã®ããã®æ·±å±¤å¦ç¿
- 第6å èªç¶è¨èªå¦çã®ããã®æ·±å±¤å¦ç¿
- 第7å ã³ã³ãã©ã¹ãã£ããã¤ãã¼ã¸ã§ã³ã¹æ³ã¨ãã®å¨è¾º
ãªã³ã¯é
- ç§ã®ããã¯ãã¼ã¯ Deep Learning
- Deep learningï¼æ·±å±¤å¦ç¿ï¼ãã¥ã¼ããªã¢ã«ãªã©
- GoogleãFacebookã注ç®ãããã£ã¼ãã©ã¼ãã³ã°ã«ã¤ãã¦ã¾ã¨ãã¦ã¿ã
- Deep Learningã®ææ - è¬ç¾©è³æ
- DeepLearningã使ã£ãå®è£ ãçºãã¦ã¿ã
解説
- DeepLearning.net - Deep Learningæ å ±ã®ç·æ¬å±±ããã¥ã¼ããªã¢ã«ãè«æãªã¹ããªã©
- Deep Learning Reading List - èªãã¹ãè«æã®ãªã¹ã
- DEEP LEARNING - Bengioããã®æ¬ã®ãã©ãã
- Neural Networks and Deep Learning - ãªã³ã©ã¤ã³ããã¯
- Learning Deep Architectures for AI
- An Introduction to Restricted Boltzmann Machines
- Unsupervised Feature Learning and Deep Learning Tutorial
- A Practical Guide to Training Restricted Boltzmann Machine
- Deep Learning: Methods and Applications
- Machine Intelligence - Natureã®ç¹é
- Deep Machine Learning - A New Frontier in Artificial Intelligence Research - ç·èª¬
- UFLDL Tutorial - ã¹ã¿ã³ãã©ã¼ã大å¦
- Deep Learning Summer School 2015
- Deep Learning (Wikipedia)
- ãã£ã¼ãã©ã¼ãã³ã°ãã¥ã¼ããªã¢ã«ï¼ãããã¯ç 究ååå ±åï¼
- ãã¥ã¼ã©ã«ãããã®é襲
- Deep Learning : Bengioå çã®ããããã¬ã·ã
- ã¯ãããDeep learning
- 人工ç¥è½ Advent Calendar 2015
Slideshare
- ä¸è¬åãã®Deep Learning
- Deep Learningæè¡ã®ä»
- Deep Learningã®åºç¤ã¨å¿ç¨äºä¾
- Deep Learningï½ä½¿ãããªãããã«ç¥ã£ã¦ãããããã¨
- ãã£ã¼ãã©ã¼ãã³ã°ã«ããç»åèªèã¨å¿ç¨äºä¾
- Deep Learningã®éå»ã¨æªæ¥ï½é»éè¡ããã®è±å´ã¸åãã¦ï½
- Deep Learningã¨ç»åèªèï½æ´å²ã»çè«ã»å®è·µï½
- Deep Learningã«ããç»åèªèé©å½ã¼æ´å²ã»ææ°çè«ããå®è·µå¿ç¨ã¾ã§ã¼
- ãã£ã¼ãã©ã¼ãã³ã°ã®åºç¤æè¡ã¨ä»å¾ã®èª²é¡ã»å±æï½ç»åèªèåéãä¸å¿ã«ï½
å½éä¼è°ãã¥ã¼ããªã¢ã«
- Representation learning tutorial - ICML2012
- Deep learning methods for vision - CVPR2012
- Deep learning for NLP (without Magic) - ACL2012
- Deep learning and its applications in signal processing - ICASSP2012
- Graduate summer school: Deep learning, feature learning
Generative Adversarial Network
- â ã¯ããã¦ã®GAN - Kerasã«ããMNISTãä¾ã«ããGAN
- â NIPS 2016 Workshop on Adversarial Training - GANã®è¬æ¼
- AdversarialNetsPapers - è«æãªã¹ã
- How to Train a GAN? Tips and tricks to make GANs work - GANãè¨ç·´ããããã®Tipé
- Chainerã使ã£ã¦ã³ã³ãã¥ã¼ã¿ã«ã¤ã©ã¹ããæããã
- Chainerã§é¡ã¤ã©ã¹ãã®èªåçæ
ã³ã³ãã¥ã¼ã¿ãã¸ã§ã³é¢é£
- â CS231n Convolutional Neural Networks for Visual Recognition - è¬ç¾©åç»ã®ãªã¹ãã¯ãã
- Building High-level Features Using Large Scale Unsupervised Learning - Googleã®ç«è«æ
- ImageNet Classification with Deep Convolutional Neural Networks
- The Shape Boltzmann Machine: A strong model of object shape
- Convolutional neural network (Wikipedia)
- äºæ¬¡å ç»åãæ¡å¤§ãããã¨æã£ããã¨ã¯ããã¾ãããï¼ - CNNã«ããè¶ è§£åãTorch7
- ããããDeepLearning - ã¢ãã¡ãã£ã©ã®é¡èªè
- ã注æã¯Deep Learningã§ããï¼ - Caffeãã¢ãã¡ãã£ã©ã®é¡èªèã
- A Neural Algorithm of Artistic Style - CNNã«ããçµµç»é¢¨ã®ç»åçæãTorch7ã«ããå®è£ ãChainerã«ããå®è£
é³å£°é¢é£
- Deep Neural Networks for Acoustic Modeling in Speech Recognition - é³å£°èªèãµã¼ãã¤ã
- Deep Learning for Acoustic Modeling in Parametric Speech Generation - é³å£°åæã»å£°è³ªå¤æãªã©ã®é³é¿ãã©ã¡ã¼ã¿çæã«ãããDNNå¿ç¨ã®ãµã¼ãã¤ã
- Composing Music With Recurrent Neural Networks · hexahedria - Recurrent Neural Networkã§ã¯ã©ã·ãã¯é³æ¥½ãçæãããããã®å¤§å¥½ãã
èªç¶è¨èªå¦çé¢é£
- èªç¶è¨èªå¦çåéã«ããããã£ã¼ãã©ã¼ãã³ã°ã®ç¾ç¶ (PDF)
- Linguistic Regularities in Continuous Space Word Representations - word2vec, King-Man+Woman=Queen
- The Unreasonable Effectiveness of Recurrent Neural Networks - RNNãç¨ããããã¹ãçæ
å¼·åå¦ç¿é¢é£
- Playing Atari with Deep Reinforcement Learning - DQN
- Human-level control through deep reinforcement learning - DQNã®Natureè«æãã³ã¼ãã¯ããã§å ¬é
- åæ£æ·±å±¤å¼·åå¦ç¿ã§ããããå¶å¾¡ - åç»ãã
- GitHub - ugo-nama-kun/DQN-chainer - Chainerã®DQNå®è£
- DQNã®çãç«ã¡ãï¼ãDeep Q-NetworkãChainerã§æ¸ãã - Pongãåç¾ã§ãã¦ãï¼ããã
Autoencoder
- Reducing the Dimensionality of Data with Neural Networks - Autoencoderãç¨ãã次å å§ç¸®ãå³4©ã¯ã¤ã³ãã¯ããããªãIsomapã¨ã©ã£ã¡ããããã ãï¼
- Ng’s Lecture Note: Sparse Autoencoder - ã¹ãã¼ã¹èªå·±ç¬¦å·åå¨ã®è§£èª¬
- æ©æ¢°å¦ç¿ãããã§ãã·ã§ãã«ã·ãªã¼ãºè¼ªèªä¼ #2 Chapter 5 ãèªå·±ç¬¦å·åå¨ã è³æ - èªå·±ç¬¦å·åå¨ã®è§£èª¬è³æã
Neural Turing Machine
Inceptionism
- Research Blog: Inceptionism: Going Deeper into Neural Networks - Google Deep Dreamãä¾ã®æ°æã¡æªãç»åã®ä½ãæ¹ã
ç 究ä¼
- æ±ç¨äººå·¥ç¥è½ã¨æè¡çç¹ç°ç¹ç ç©¶ä¼ - å ¨è³ã¢ã¼ããã¯ãã£ãDeep Learningé¢é£ã®è©±é¡ãå¤ã
- Deep Learning Japan - æ¾å°¾ããã°ã«ã¼ãããªã³ã¯éããã
Theano
- Theano - Deep Learningã®å®è£ ã«é©ããæ°å¤è¨ç®ã©ã¤ãã©ãª
- Theano Tutorial - Theanoã®ãã¥ã¼ããªã¢ã«
- Deep Learning Tutorial - Theanoãç¨ããDeep Learningã¢ã«ã´ãªãºã å®è£ ã®è§£èª¬
- Theanoå ¥é
- Theano ã® åºæ¬ã¡ã¢
- Theano解説
- Theanoã§Deep Learning
- Deep Learningãå®è£ ãã
- Theano Tutorial - IPython Notebookå½¢å¼ã®ãã¥ã¼ããªã¢ã«
- GitHub - JonathanRaiman/theano_lstm: Nano size Theano LSTM module - Theanoã®LSTMå®è£
- Pylearn2 - Theanoã使ã£ãDeep Learningã©ã¤ãã©ãª
- Lasagne - Theanoã®ã©ããã¼ã©ã¤ãã©ãª
- nolearn - Theanoãã¼ã¹ï¼
- keras - Theanoã¨Tensorflowãã¼ã¹ãå®è£ ãã·ã³ãã«ã§ã¨ã¦ãç¾ãããæè¿ã¯ã¾ã£ã¦ããã
Caffe
- Caffe - ã³ã³ãã¥ã¼ã¿ãã¸ã§ã³åéã§ãã使ããã¦ããããã
- Caffeã§æ軽ã«ç»ååé¡
- Caffeã§Deep Q-Networkãå®è£ ãã¦æ·±å±¤å¼·åå¦ç¿ãã¦ã¿ã
Chainer
- Chainer - Preferred Networksã
- Deep Learning ã®ãã¬ã¼ã ã¯ã¼ã¯ Chainer ãå ¬éãã¾ãã
- ãã£ã¼ãã©ã¼ãã³ã° ãã¬ã¼ã ã¯ã¼ã¯Chainerã試ããªãã解説ãã¦ã¿ã
- æ©æ¢°å¦ç¿ã©ã¤ãã©ãª Chainerã®ç´¹ä»
- Chainerã§CIFAR-10ã®åé¡ãè¡ã£ã¦ã¿ã
- Chainerã§Autoencoderã試ãã¦çµæãå¯è¦åãã¦ã¿ã
- æ©æ¢°å¦ç¿åå¿è ã1ããChainerã®ã³ã¼ããæ¸ãå§ããããã¾ã§ã«å¿ è¦ã ã£ããã¨ãã¾ã¨ãã
TensorFlow
- TensorFlow - Google
- TensorFlow - Googleâs latest machine learning system, open sourced for everyone
ãã®ä»ã©ã¤ãã©ãª
- Deeplearning4j - Javaã©ã¤ãã©ãª
- H2O
- Torch7 - Facebook
MOOCã»åç»
- Machine Learning - Ngããã«ããæ©æ¢°å¦ç¿è¬ç¾©ãæ¥æ¬èªåå¹ãããDeep Learningã¯æ±ã£ã¦ãã¾ãããããã¸ã¹ãã£ãã¯å帰ããã¥ã¼ã©ã«ãããã®æè¡ãå®è·µçã«å¦ã¹ãããããããã
- Neural Networks for Machine Learning - Hintonããã«ãããã¥ã¼ã©ã«ãããè¬ç¾©
- The Next Generation of Neural Networks - Google Talksã§ã®è¬æ¼
- Recent Developments in Deep Learning - Google Talksã§ã®è¬æ¼
- Introduction to Deep Learning with Python
- Deep Learning & Neural Networks short course
- Unsupervised Feature Learning and Deep Learning
- DeepLearning.TV
- Deep Learning Summer School 2015 - æè¿ã¯ããè¦ã¦ã
æ¸ç±
- 人工ç¥è½ã¯äººéãè¶ ããã (è§å·ï¼¥ï¼°ï¼µï¼¢é¸æ¸) - ãªãDeep Learningããããªã«é¨ããã¦ããã®ãç¥ããã人åãã
- [asin:B00OH3ZU7Y:title]
- AIã®è¡æã人工ç¥è½ã¯äººé¡ã®æµã (è¬è«ç¤¾ç¾ä»£æ°æ¸)
- ãªã³ã©ã¤ã³æ©æ¢°å¦ç¿ (æ©æ¢°å¦ç¿ãããã§ãã·ã§ãã«ã·ãªã¼ãº) - å¾é æ³ã®èª¬æãåèã«ãªãã
- 深層å¦ç¿ (æ©æ¢°å¦ç¿ãããã§ãã·ã§ãã«ã·ãªã¼ãº) - è¦ç¹ãç°¡æ½ã«ã¾ã¨ã¾ã£ã¦ãã¦ããã£ãã
- ããªã¼ã½ããã§ã¯ãããæ©æ¢°å¦ç¿å ¥é
- ãã£ã¼ãã©ã¼ãã³ã°ãããã°ãã¼ã¿ãæ©æ¢°å¦ç¿ãããããã¯ãã®å¿çå¦
- Machine Learning: A Probabilistic Perspective (Adaptive Computation and Machine Learning series) - æå¾ã®ç« ã«Deep Learningã®èª¬æããã
- Deep Learning: A Practitioner's Approach
- Fundamentals of Deep Learning: Designing Next-generation Machine Intelligence Algorithms
- Learning Deep Architectures for AI (Foundations and Trends(r) in Machine Learning) - ãã®è«æã¨åãã¿ã¤ãã«ã ãæ§æãéãã¿ãã
- Deep Learning: Methods and Applications (Foundations and Trends in Signal Processing)
- ã³ã³ãã¥ã¼ã¿ãã¸ã§ã³æå 端ã¬ã¤ã6 (CVIMãã¥ã¼ããªã¢ã«ã·ãªã¼ãº) - “ãã£ã¼ãã©ã¼ãã³ã°ã®è§£èª¬ãç§é¸"ã¨ã®ãã¨
- WEB+DB PRESS Vol.89 - Chainerã«ãã深層å¦ç¿ã®ç¹éè¨äº
- 深層å¦ç¿ Deep Learning (ç£ä¿®:人工ç¥è½å¦ä¼) - 人工ç¥è½å¦ä¼ã®ç¹éè¨äºãã¾ã¨ããæ¸ç±
- ã¼ãããä½ãDeep Learning âPythonã§å¦ã¶ãã£ã¼ãã©ã¼ãã³ã°ã®çè«ã¨å®è£