3 Credit Hour Course
Intended For Level 0 Term 0 Students
Prerequisite:
Reinforcement Learning (RL) task formulation: action space, state space, environment definition; Markov decision processes; Tabular based solutions: dynamic programming, Monte Carlo, temporal-difference; Function approximation solutions: Deep Q-networks; Model-free RL: policy gradients, proximal policy optimization, deep deterministic policy gradient, actor-critic algorithms, value function methods, RL with Q-functions; Model-based reinforcement learning: stochastic optimization, Monte Carlo tree search, trajectory optimization, uncertainty in model-based RL, model-based policy learning; Imitation learning: behavioral cloning, inverse RL, generative adversarial imitation learning; Transfer and multitask learning; Meta reinforcement learning; Distributed reinforcement learning; Multi-agent learning: partial observable environments; Bandits, exploration and exploitation, Hierarchical RL; Applications of RL in robotics, autonomous systems, natural language processing, and game theory.
