Lunar Lander: Reinforcement Learning Using Deep Q-Learning

This project is about Lunar Lander, an OpenAI Gym environment where a rocketship has to learn how to land on its own via exploration. There are 8 states and 4 actions. Rewards are based on actions taken and are pre-defined. The goal is to obtain an average of 200 points over 100 episodes after training. This can be solved using Q-learning with a neural network as the function approximator.

More info to be added…