23b3003_SOC

• Project Name : General Alpha Zero
• Project ID : 16
• Mentee : Vaibhav Verma
• Mentor : Aman Moon

Overview

The General AlphaZero project aims to create a sophisticated, self-improving game-playing AI inspired by the groundbreaking research from Google DeepMind, titled "Mastering Chess and Shogi by Self-Play with General Reinforcement Learning Algorithm". This project teaches an AI to play games autonomously, improving its skills through continuous self-play and learning. The core technology is a modified version of the Monte Carlo Tree Search (MCTS) algorithm, which allows the AI to explore various game scenarios and learn from them. The AI refines its strategies using Reinforcement Learning and is powered by Residual Networks (ResNets) as the backbone of its neural network architecture.

Repository Structure

• GeneralAlphaZero.ipynb: The main project file implementing the AlphaZero algorithm.
• Pytorch/: Contains all the PyTorch-related files used during development and learning.
• CNN_Pytorch.py: Implementation of Convolutional Neural Networks (CNNs) using PyTorch.
• ResNets_Pytorch.py: Implementation of Residual Networks (ResNets) using PyTorch.
• NNfromScratch.ipynb: Implementation of neural networks from scratch, without using any external libraries.