SUMMARY.md

Summary

Introduction to Deep RL

• Introduction
• Review
  • Reinforcement Learning
  • Deep Learning
• Why is deep learning useful for RL?
• Challenges of Deep RL
• Landscape of methods

Value-Based Methods

• Value Function Estimation
• Action-Value Function Estimation
• Neural-fitted Q iteration (NFQ)
• Deep Q Networks (DQN)
• DQN Extensions
  • Double DQN
  • Dueling DQN
  • Prioritized Experience Replay
  • Normalized Advantage Function
  • Bootstrapped DQN
• Applications
  • TD-Gammon (1992)

Policy Search

• Introduction
• The REINFORCE gradient estimator
• Policy Gradient Extensions
  • Trust Region Policy Optimization
  • Natural Policy Gradients
  • Proximal Policy Optimization
• Actor-Critic Paradigm
• Distributed RL
  • Gorila
• Advantage Actor-Critic (A2C)
• Applications
  • Robotics

Model Estimation

• Introduction
• Monte Carlo Tree Search (MCTS)
• UCT-to-Realtime
• Applications
  • AlphaGo (2017)

Hierarchical RL

• Introduction to Hierarchical RL
• Options
• Options DQN
• Option-Critic Architecture
• Hierarchical DQN
• Feudal Networks

Inverse RL

Multi-Agent RL