[
  {
    "path": "README.md",
    "content": "# Some resources for ML research\n\nPersonal and biased selection of ML resources.\n\n**Disclaimer:** I'm a noivce in ML research, and I read only a few of the list.\n\n\n## Table of Contents\n- [Beginner's Guide](#beginners-guide)\n- [Machine Learning](#machine-learning)\n- [Deep Learning](#deep-learning)\n- [Generative Model](#generative-model)\n- [Reinforcement Learning](#reinforcement-learning)\n- [Graphical Model](#graphical-model)\n- [Optimization](#optimization)\n- [Learning Theory](#learning-theory)\n- [Statistics](#statistics)\n- [Topics in Machine Learning](#topics-in-machine-learning)\n- [Math Backgrounds](#math-backgrounds)\n- [Blogs](#blogs)\n\n\n## Beginner's Guide\n\n**Must Read**\n- Machine Learning: A Probabilistic Perspective (Murphy)\n- Deep Learning (Goodfellow et al.)\n- Reinforcement Learning: An Introduction (Sutton & Barto)\n\n**Recommended**\n- Convex Optimization (Boyd & Vandenberghe)\n- Graphical Models, Exponential Families, and Variational Inference (Wainwright & Jordan)\n- Learning from Data (Abu-Mostafa) *-> for whom interested in learning theory*\n\n**Recent Topics**\n- Read research blogs (e.g., [OpenAI](https://blog.openai.com/), [BAIR](http://bair.berkeley.edu/blog/), [CMU](https://blog.ml.cmu.edu/))\n- Read lectures from Berkeley, Stanford, CMU or UofT (e.g., [unsupervised learning](https://sites.google.com/view/berkeley-cs294-158-sp19))\n- There are lots of good sources, but I stop updating them up-to-date\n\n\n## Machine Learning\nThere are many ML books, but most of them are encyclopedic. <br/>\nI recommend to take a course using Murphy or Bishop book.\n\n### Textbook\n- Machine Learning: A Probabilistic Perspective (Murphy) :sparkles:\n- Pattern Recognition and Machine Learning (Bishop) :sparkles:\n- The Elements of Statistical Learning (Hastie et al.)\n- Pattern Classification (Duda et al.)\n- Bayesian Reasoning and Machine Learning (Barber)\n\n### Lecture\n- [Stanford CS229 Machine Learning](http://cs229.stanford.edu) :sparkles:\n- [CMU 10701 Mahine Learning](http://www.cs.cmu.edu/~tom/10701_sp11/)\n- [CMU 10702 Statistical Machine Learning](http://www.stat.cmu.edu/~larry/=sml/)\n- [Oxford Machine Learning](https://www.cs.ox.ac.uk/people/nando.defreitas/machinelearning/)\n\n\n## Deep Learning\nGoodfellow et al. is the new classic. <br/>\nFor vision and NLP, Stanford lectures would be helpful.\n\n### Textbook\n- Deep Learning (Goodfellow et al.) :sparkles:\n\n### Lecture (Practice)\n- [Deep Learning book](http://www.deeplearningbook.org/lecture_slides.html) :sparkles:\n- [Stanford CS231n Convolutional Neural Networks for Visual Recognition](http://cs231n.stanford.edu) :sparkles:\n- [Stanfrod CS224d Deep Learning for Natural Language Processing](http://cs224d.stanford.edu)\n- [Stanfrod CS224s Spoken Language Processing](http://web.stanford.edu/class/cs224s/)\n- [Oxford Deep Natural Language Processing](https://github.com/oxford-cs-deepnlp-2017/lectures)\n- [CMU 11747 Neural Networks for NLP](http://phontron.com/class/nn4nlp2017/index.html)\n\n### Lecture (Theory)\n- [Stanford Stat385 Theories of Deep Learning](https://stats385.github.io/)\n\n\n## Generative Model\nI seperated generative model as an independent topic, <br/>\nsince I think it is big and important area.\n\n### Lecture\n- [Toronto CSC2541 Differentiable Inference and Generative Models](https://www.cs.toronto.edu/~duvenaud/courses/csc2541/index.html)\n- [Toronto CSC2547 Learning Discrete Latent Structure](https://duvenaud.github.io/learn-discrete/)\n- [Toronto CSC2541 Scalable and Flexible Models of Uncertainty](https://csc2541-f17.github.io/)\n\n\n## Reinforcement Learning\nFor classic (non-deep) RL, Sutton & Barto is the classic. <br/>\nFor deep RL, lectures from Berkeley/CMU looks good.\n\n### Textbook\n- Reinforcement Learning: An Introduction (Sutton & Barto) :sparkles:\n\n### Lecture\n- [UCL Reinforcement Learning](http://www0.cs.ucl.ac.uk/staff/d.silver/web/Teaching.html) :sparkles:\n- [Berkeley CS294 Deep Reinforcement Leanring](http://rll.berkeley.edu/deeprlcourse/) :sparkles:\n- [CMU 10703 Deep Reinforcement Learing and Control](https://katefvision.github.io/)\n\n\n## Graphical Model\nKoller & Friedman is comprehensive, but too encyclopedic. <br/>\nI recommend to take an introductory course using Koller & Friedman book. <br/>\n\nWainwright & Jordan only focuses on variational inference, <br/>\nbut it gives really good intuition for probabilistic models.\n\n### Textbook\n- Probabilistic Graphical Models: Principles and Techniques (Koller & Friedman)\n- Graphical Models, Exponential Families, and Variational Inference (Wainwright & Jordan) :sparkles:\n\n### Lecture\n- [Stanford CS228 Probabilistic Graphical Models](http://cs.stanford.edu/~ermon/cs228/index.html)\n- [CMU 10708 Probabilistic Graphical Models](http://www.cs.cmu.edu/~epxing/Class/10708/) :sparkles:\n\n\n## Optimization\nBoyd & Vandenberghe is the classic, but I think it's too boring. <br/>\nReading chapter 2-5 would be enough.\n\nBertsekas more concentrates on convex analysis. <br/>\nNocedal & Wright more concentrates on optimization.\n\n### Textbook\n- Convex Optimization (Boyd & Vandenberghe) :sparkles:\n- Convex Optimization Theory (Bertsekas)\n- Numerical Optimization (Nocedal & Wright)\n\n### Lecture\n- [Stanford EE364a Convex Optimization I](http://stanford.edu/class/ee364a/) :sparkles:\n- [Stanford EE364b Convex Optimization II](http://stanford.edu/class/ee364a/)\n- [MIT 6.253 Convex Analysis and Optimization](https://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-253-convex-analysis-and-optimization-spring-2012/index.htm)\n\n\n## Learning Theory\nIn my understanding, there are two major topics in learning theory:\n\n- **Learning Theory:** VC-dimension, PAC-learning\n- **Online Learning:** regret bound, multi-armed bandit\n\nFor learning theory, Kearns & Vazirani is the classic; but it's too old-fashined. <br/>\nAbu-Mostafa is a good introductory book, and I think it's enough for most people.\n\nFor online learning, Cesa-Bianchi & Lugosi is the classic. <br/>\nFor multi-armed bandit, Bubeck & Cesa-Bianchi provides a good survey.\n\n### Textbook (Learning Theory)\n- Learning from Data (Abu-Mostafa) :sparkles:\n- Foundations of Machine Learning (Mohri et al.)\n- An Introduction to Computational Learning Theory (Kearns & Vazirani) \n\n### Textbook (Online Learning)\n- Prediction, Learning, and Games (Cesa-Bianchi & Lugosi)\n- Regret Analysis of Stochastic and Nonstochastic Multi-armed Bandit Problems (Bubeck & Cesa-Bianchi)\n\n### Lecture\n- [Caltech Learning from Data](https://work.caltech.edu/telecourse.html) :sparkles:\n- [CMU 15859 Machine Learning Theory](http://www.cs.cmu.edu/~avrim/ML14/)\n- [Berkeley CS281b/Stat241b Statistical Learning Theory](https://www.stat.berkeley.edu/~bartlett/courses/2014spring-cs281bstat241b/)\n- [MIT 9.520 Statistical Learning Theory and Applications](http://www.mit.edu/~9.520/fall15/)\n\n\n## Statistics\nStatistics is a broad area; hence, I listed only a few of them. <br/>\nFor advanced topics, lectures from Berkeley/Stanford/CMU/MIT looks really cool. <br/>\n\n### Textbook (Statistical Inference)\n- All of Statistics (Wasserman)\n- Computer Age Statistical Inference (Efron & Hastie) :sparkles:\n- Time Series Analysis and Its Applications: With R Examples (Shumway & Stoffer)\n\n### Textbook (Nonparametrics)\n- All of Nonparametric Statistics (Wasserman)\n- Introduction to Nonparametric Estimation (Tsybakov)\n- Gaussian Process and Machine Learning (Rasmussen & Williams) :sparkles:\n- Bayesian Nonparametrics (Ghosh & Ramamoorthi) :sparkles:\n\n### Textbook (Advanced Topics)\n- High-Dimensional Statistics: A Non-Asymptotic Viewpoint (Wainwright) :sparkles:\n- Statistics for High-Dimensional Data (Bühlmann & van de Geer)\n- Asymptotic Statistics (van der Vaart)\n- Empirical Processes in M-Estimation (van der Vaart)\n\n### Lecture\n- [Berkeley Stat210a Theoretical Statistics I](https://www.stat.berkeley.edu/~wfithian/courses/stat210a/)\n- [Berkeley Stat210b Theoretical Statistics II](https://people.eecs.berkeley.edu/~jordan/courses/210B-spring17/)\n- [Stanford Stat300a Theory of Statistics](https://web.stanford.edu/~lmackey/stats300a/)\n- [Stanford CS369m Algorithms for Massive Data Set Analysis](http://cs.stanford.edu/people/mmahoney/cs369m/)\n- [CMU 36755 Advanced Statistical Theory I](http://www.stat.cmu.edu/~arinaldo/36755/F16/)\n- [MIT 18.S997 High-Dimensional Statistics](https://ocw.mit.edu/courses/mathematics/18-s997-high-dimensional-statistics-spring-2015/)\n\n\n## Topics in Machine Learning\nMiscellaneous topics related to machine learning. <br/>\nThere are much more subfields, but I'll not list them all.\n\n### Information Theory\n- Elements of Information Theory (Cover & Thomas)\n- Information Theory, Inference, and Learning Algorithms (MacKay)\n\n### Network Science\n- Networks, Crowds, and Markets (Easley & Kleinberg)\n- Social and Economic Networks (Jackson)\n\n### Markov Chain\n- Markov Chains (Norris)\n- Markov Chains and Mixing Times (Levin et al.)\n\n### Game Theory\n- Algorithmic Game Theory (Nisan et al.)\n- Multiagent Systems (Shoham & Leyton-Brown)\n\n### Combinatorics\n- The Probabilistic Method (Alon & Spencer)\n- A First Course in Combinatorial Optimization (Lee)\n\n### Algorithm\n- Introduction to Algorithms (Cormen et al.)\n- Randomized Algorithms (Motwani & Raghavan)\n- Approximation Algorithms (Vazirani)\n\n### Geometric View\n- Topological Data Analysis (Wasserman)\n- Methods of Information Geometry (Amari & Nagaoka)\n- Algebraic Geometry and Statistical Learning Theory (Watanabe)\n\n### Some Lectures\n- [MIT 18.409 Algorithmic Aspects of Machine Learning](http://people.csail.mit.edu/moitra/409.html)\n- [MIT 18.409 An Algorithmist's Toolkit](http://stellar.mit.edu/S/course/18/fa09/18.409/)\n\n\n## Math Backgrounds\nI selected essential topics for machine learning. <br/>\nPersonally, I think more analysis / matrix / geometry never hurts.\n\n### Probability\n- Probability: Theory and Examples (Durrett)\n- Theoretical Statistics (Keener)\n- Stochastic Processes (Bass)\n- Probability and Statistics Cookbook (Vallentin)\n\n### Linear Algebra\n- Linear Algebra (Hoffman & Kunze)\n- Matrix Analysis (Horn & Johnson)\n- Matrix Computations (Golub & Van Loan)\n- The Matrix Cookbook (Petersen & Pedersen)\n\n### Large Deviations\n- Concentration Inequalities and Martingale Inequalities (Chung & Lu)\n- An Introduction to Matrix Concentration Inequalities (Tropp)\n\n\n## Blogs\n\n- [Google AI Blog](https://ai.googleblog.com/)\n- [DeepMind Blog](https://deepmind.com/blog/?category=research)\n- [OpenAI Blog](https://blog.openai.com/)\n- [FAIR Blog](https://research.fb.com/blog/)\n- [Distill.pub](https://distill.pub/)\n- [BAIR Blog](http://bair.berkeley.edu/blog/)\n- [CMU Blog](https://blog.ml.cmu.edu/)\n- [Off the convex path](http://www.offconvex.org/)\n- [inFERENCe](http://www.inference.vc/)\n- [Sebastian Ruder](http://ruder.io/#open)\n- [Lunit Tech Blog (Korean)](https://blog.lunit.io/category/paper-review/)\n"
  }
]