Deep Reinforcement Learning with PyTorch (Super Mario Bros)

Deep Learning
AI
PyTorch
Python
Machine Learning
Algorithm
Games
Author

David Munoz Tord

Published

February 4, 2022

Reinforcement Learning with PyTorch (Super Mario Bros)

Here is a project I worked on to train AI agents to play and solve games using reinforcement learning. The project is based on the Super Mario Bros game, and the AI agents are trained using the Proximal Policy Optimization (PPO) algorithm. The agents are implemented using the PyTorch library, and the game environment is provided by the OpenAI Gym library . The project also includes a custom implementation of the game environment, which is based on the Super Mario Bros game for the Nintendo Entertainment System (NES). The game environment is implemented using the FCEUX emulator, and the game is controlled using the Python library PyAutoGUI.

Super Mario Bros

I also used TensorBoard to visualize and evaluate the quality metrics of the model.

TensorBoard

Check out the project on GitHub.