Training Q-learning agents 🤖
in OpenAI-Gymnasium

1. Frozen Lake 🥶

Frozen lake involves crossing a frozen lake from start to goal without falling into any holes by walking over the frozen lake. The player may not always move in the intended direction due to the slippery nature of the frozen lake.

Training plots

Training Q-Learning agent on the FrozenLake-v1 environment.

Results

Untrained Agent	Trained Agent

2. Space Invaders 👾

Results

Acknowledgements 🫡

• Code inspired from Deep Reinforcement Learning Course by Thomas Simonini (http://simoninithomas.com/)
• David Silver’s course in particular lesson 4 and lesson 5
• Reinforcement Learning Specialization from Coursera