Reinforcement Learning Agent

An implementation of a reinforcement learning agent using Q-learning with epsilon-greedy exploration. The agent learns optimal policies in discrete state-action spaces through iterative interaction with simulated environments. The project includes implementations of tabular Q-learning, Deep Q-Networks (DQN) with experience replay, and double DQN variants. Training visualizations show convergence curves, Q-value heatmaps, and policy evolution over episodes. The codebase is structured as an educational resource with detailed docstrings explaining the theory behind each algorithm component.