DelGo

Training a computer (and myself) to play Go!

Note: This project is a WIP— cleaning up the code piece by piece and adding it to GitHub

New this Version:

• Using Monte-Carlo Tree Search for selecting better than random moves
• Can customize board size and player order!

Currently Implemented:

• Outlining the rules, board, and game logic
• Zobrist Hashing to speed up gameplay
• Creating a bot that can play against itself and other humans
• Simple sequential network used for move prediction (not the best, but better than random)
• Implemented Convolutional Neural Networks + New Activation Functions (Softmax & Rectified Linear Units)!
• Added dropout layers to prevent overfitting, and using Categorial Cross-Entropy instead of MSE for accuracy measuring

Currently Working On:

• Deep Learning using game records from high level players
• A Better GUI

Based off the book Deep Learning and the Game of Go by Max Pumperla and Kevin Ferguson