StockfishNN is a modified version of the Stockfish chess engine, integrating neural network-based heuristics to enhance its decision-making process. This project aims to explore the potential of combining traditional minimax search strategies with the adaptive insights provided by deep neural networks.

StockfishNN emerges from a study comparing the search algorithms of Stockfish and Leela Chess Zero (Lc0), focusing on their performance in atypical game states. The findings indicated a potential advantage in blending the exhaustive search capabilities of Stockfish with the heuristic intuition of neural networks. StockfishNN represents an effort to harness this synergy, aiming to improve performance in diverse chess scenarios.

• Integration of neural network heuristics with Stockfish's search algorithm.
• Aimed at enhancing adaptability and performance in varied chess positions.
• Developed through extensive research and simulation.

StockfishNN operates similarly to the original Stockfish engine. Users familiar with Stockfish should find the transition to StockfishNN straightforward. For instructions on using StockfishNN, including configuration and integration, please refer to the official Stockfish website.

StockfishNN is based on the official Stockfish engine. We acknowledge the Stockfish community for providing the foundation upon which this project is built.

Contributions to StockfishNN are welcome. Whether you have suggestions, improvements, or want to collaborate on research, your input is valued. Please feel free to submit pull requests or reach out with your ideas.