Pong_Reinforcememnt_Learning_Policy_Gradients

I have tried explaining the most simple reinforcement learning algortithm Policy Gradients to make a game bot for playing Pong Game Try my colab notebook to get started quickly.

Installation

Use the package manager pip to install OpenAI's Gym Library.

pip install gym

Install pickle library

pip install pickle-mixin

Dependencies

numpy
matplotlib
pickle

Usage

python pong_.py

Example of Learned moves

Learned weights of complete first layer to first node of second layer

Credits

The credits for this code go to Andrej Karpathy and his famous blog post.I've merely created a wrapper to get people started and explore this domain.

What's Next?

I have went ahead and made a two playered Pong Game where both players are AI that are learning by playing with each other.. Do check it out here!