Deep Pepper: Expert Iteration based Chess agent in the Reinforcement Learning Setting

TL;DR

Deep Pepper introduces a reinforcement learning chess agent that leverages embedded knowledge to accelerate training, providing a faster alternative to tabula rasa systems like Alpha Zero, and includes detailed methodology and future directions.

Contribution

The paper presents a novel chess agent that incorporates embedded knowledge to improve training efficiency in reinforcement learning, along with detailed algorithmic insights and open-source code.

Findings

Faster training of chess agents using embedded knowledge

Mathematical formulation of the algorithm

Potential future research directions

Abstract

An almost-perfect chess playing agent has been a long standing challenge in the field of Artificial Intelligence. Some of the recent advances demonstrate we are approaching that goal. In this project, we provide methods for faster training of self-play style algorithms, mathematical details of the algorithm used, various potential future directions, and discuss most of the relevant work in the area of computer chess. Deep Pepper uses embedded knowledge to accelerate the training of the chess engine over a "tabula rasa" system such as Alpha Zero. We also release our code to promote further research.