yandexdataschool/Practical_RL

Practical_RL

An open course on reinforcement learning in the wild. Taught on-campus at HSE and YSDA and maintained to be friendly to online students (both english and russian).

Manifesto:

• Optimize for the curious. For all the materials that aren’t covered in detail there are links to more information and related materials (D.Silver/Sutton/blogs/whatever). Assignments will have bonus sections if you want to dig deeper.
• Practicality first. Everything essential to solving reinforcement learning problems is worth mentioning. We won't shun away from covering tricks and heuristics. For every major idea there should be a lab that makes you to “feel” it on a practical problem.
• Git-course. Know a way to make the course better? Noticed a typo in a formula? Found a useful link? Made the code more readable? Made a version for alternative framework? You're awesome! Pull-request it!

Course info

• FAQ: About the course, Technical issues thread, Lecture Slides, Online Student Survival Guide

• Anonymous feedback form.

• Virtual course environment:

  • Google Colab - set open -> github -> yandexdataschool/pracical_rl -> {branch name} and select any notebook you want.
  • Installing dependencies on your local machine (recommended).
  • Alternative: Azure Notebooks.

Additional materials

• RL reading group

Syllabus

The syllabus is approximate: the lectures may occur in a slightly different order and some topics may end up taking two weeks.

• week01_intro Introduction

  • Lecture: RL problems around us. Decision processes. Stochastic optimization, Crossentropy method. Parameter space search vs action space search.
  • Seminar: Welcome into openai gym. Tabular CEM for Taxi-v0, deep CEM for box2d environments.
  • Homework description - see week1/README.md.

• week02_value_based Value-based methods

  • Lecture: Discounted reward MDP. Value-based approach. Value iteration. Policy iteration. Discounted reward fails.
  • Seminar: Value iteration.
  • Homework description - see week2/README.md.

• week03_model_free Model-free reinforcement learning

  • Lecture: Q-learning. SARSA. Off-policy Vs on-policy algorithms. N-step algorithms. TD(Lambda).
  • Seminar: Qlearning Vs SARSA Vs Expected Value SARSA
  • Homework description - see week3/README.md.

• recap_deep_learning - deep learning recap

  • Lecture: Deep learning 101
  • Seminar: Intro to pytorch/tensorflow, simple image classification with convnets

• week04_approx_rl Approximate (deep) RL

  • Lecture: Infinite/continuous state space. Value function approximation. Convergence conditions. Multiple agents trick; experience replay, target networks, double/dueling/bootstrap DQN, etc.
  • Seminar: Approximate Q-learning with experience replay. (CartPole, Atari)

• week05_explore Exploration

  • Lecture: Contextual bandits. Thompson Sampling, UCB, bayesian UCB. Exploration in model-based RL, MCTS. "Deep" heuristics for exploration.
  • Seminar: bayesian exploration for contextual bandits. UCB for MCTS.

• week06_policy_based Policy Gradient methods

  • Lecture: Motivation for policy-based, policy gradient, logderivative trick, REINFORCE/crossentropy method, variance reduction(baseline), advantage actor-critic (incl. GAE)
  • Seminar: REINFORCE, advantage actor-critic

• week07_seq2seq Reinforcement Learning for Sequence Models

  • Lecture: Problems with sequential data. Recurrent neural networks. Backprop through time. Vanishing & exploding gradients. LSTM, GRU. Gradient clipping
  • Seminar: character-level RNN language model

• week08_pomdp Partially Observed MDP

  • Lecture: POMDP intro. POMDP learning (agents with memory). POMDP planning (POMCP, etc)
  • Seminar: Deep kung-fu & doom with recurrent A3C and DRQN

• week09_policy_II Advanced policy-based methods

  • Lecture: Trust region policy optimization. NPO/PPO. Deterministic policy gradient. DDPG
  • Seminar: Approximate TRPO for simple robot control.

• week10_planning Model-based RL & Co

  • Lecture: Model-Based RL, Planning in General, Imitation Learning and Inverse Reinforcement Learning
  • Seminar: MCTS for toy tasks

• yet_another_week Inverse RL and Imitation Learning

  • All that cool RL stuff that you won't learn from this course :)

Course staff

Course materials and teaching by: [unordered]

• Pavel Shvechikov - lectures, seminars, hw checkups, reading group
• Nikita Putintsev - seminars, hw checkups, organizing our hot mess
• Alexander Fritsler - lectures, seminars, hw checkups
• Oleg Vasilev - seminars, hw checkups, technical support
• Dmitry Nikulin - tons of fixes, far and wide
• Mikhail Konobeev - seminars, hw checkups
• Ivan Kharitonov - seminars, hw checkups
• Ravil Khisamov - seminars, hw checkups
• Anna Klepova - hw checkups
• Fedor Ratnikov - admin stuff

Contributions

• Using pictures from Berkeley AI course
• Massively refering to CS294
• Several tensorflow assignments by Scitator
• A lot of fixes from arogozhnikov
• Other awesome people: see github contributors
• Alexey Umnov helped us a lot during spring2018