Reinforcement Learning

Solutions for known reinforcement learning problems following Barto and Sutton's second edition book, taught at an advanced Reinforcement Learning course. This includes problems like Multi-Armed Bandits, Gambler's Problem, The Cliff Problem, and TD-Learning with Function Approximation.

Homework 1

• h1.pdf
• h1_sol.pdf

Multi-Armed Bandits Problem

• multi-armed_bandit_hw_1_1.ipynb
  • Greedy Policy
  • Epsilon Greedy Policy
  • UCB Policy

Gambler's Problem

• gamblers_problem_hw_1_2.ipynb

Homework 2

• h2.pdf
• h2_sol.pdf

The Cliff Problem

• cliff_problem_hw_2_1.ipynb

TD-Learning with Function Approximation

• td_learning_function_approximation_hw_2_2.ipynb