This notebook has been created by Alessandro Pomponio for the 2021 Autonomous and Adaptive System course, held by prof. Mirco Musolesi at the University of Bologna. It features an autonomous agent that learns to play the Atari Bowling game using Reinforcement Learning and Proximal Policy Optimization.
The repository contains the Jupyter notebook used to train the agent (training time was ~36h on a quad-core VPS, cpu-only) and a report.
Bowling.111.mp4
Bowling.strikes.mp4
The code might contain errors, be wary of that!