Chess Bot
This project involved developing a Python-based chess bot that uses advanced search algorithms and neural networks to predict reasonable moves and challenge human players.
Implementation Details
- Programming Language: Python
- Frameworks and Libraries: TensorFlow, PyTorch
- Testing Platform: Chess.com for detailed analysis and benchmarking
Used data
The chess bot was trained on a large dataset of chess positions and moves sourced from online chess databases. The data was preprocessed for use in neural networks, allowing the bot to learn and improve its gameplay.
Algorithm
The chess bot uses the MinMax algorithm with a heuristic based on figure values. This approach evaluates possible moves by considering the value of each chess piece and the positional advantage.
Testing and Evaluation
Extensive testing was conducted by playing matches against other chess programs and human players. Performance metrics were gathered to compare different prototypes, with the best model achieving an ELO rating capable of challenging average human players.
Results
Through iterative development and testing, the chess bot showed significant improvement in its playing ability. The final prototype successfully defeated programs with an ELO rating of 1200 and performed well against human opponents, demonstrating its capability to predict logical and competitive moves.