aurelien-renault / RL_Rainbow

A analysis and test of the Rainbow agent for reinforcement learning

RL Rainbow agent

Overview

This school project goal was to analyse the paper "Rainbow : Combining Improvements in Deep Reinforcement Learning" talking about several Deep Q-Network extensions and the combination of them into one agent, Rainbow.

Collaboration

This project have been done in collaboration by Ikhlass Yaya-Oyé, Aurélien Renault and Victor Clermont.

You will find in this repositery :

• A LaTeX report of our study
• The test video of each agent
• The algorithms implementation