Deep Reinforcement Learning Project (Competition)

This Deep Reinforcement Learning Project was prepared by:

1. Lang Kah Chun
2. Yee Qing Wei
3. Sun Qi Yang

Final Project

The objective of this final project is to train a smart agent to play the Atari Breakout-V5 game and conduct a comparative analysis of different algorithms used in Deep Reinforcement Learning (DeepRL).

Competition

• Rainbow Algorithm: We successfully implemented the Rainbow algorithm by referring to this repository.
• Class Competition: We used the trained agent to compete in a class competition environment for the Atari Pong game, as detailed in this repository.

Summary

This project demonstrates the application of advanced Deep Reinforcement Learning techniques to train agents capable of playing classic Atari games and reviews the performance of each individual algorithm's strengths and weaknesses through comparative analysis.