Reinforcement Learning

Implementation of various RL agents

This project provides a collection of reinforcement learning agents and offers a unified framework to train, evaluate and visualize the performance of the agents on various Gym environments.

The following agents have been implemented:

  • Tabular learning
    • Q-Learning QLearningAgent
    • SARSA Learning SARSAAgent
  • Deep-Learning based function approximation
    • Dueling Deep Q-Network DDQNAgent
    • REINFORCE algorithm ReinforceMCwithBaselineAgent, ReinforceMCwithoutBaselineAgent
  • Hierarchical Learning
    • Semi-Markov Decision Process Q-Learning SMDPQLearningAgent
    • Intra-Option Q-Learning IntraOptionQLearningAgent

More information about this project can be found at this repository