This repository features an implementation of the Double Deep Q-Network (Double DQN) reinforcement learning algorithm to master the game of Tic-Tac-Toe.
The project utilizes Deep Reinforcement Learning to train an agent that can make optimal moves in a competitive environment. The use of Double DQN helps in reducing the overestimation of Q-values, leading to more stable training.
- Double DQN Agent: A neural-network-based agent trained using reinforcement learning.
- Configurable Environment: A flexible game environment for training and evaluation.
- Reproducible Experiments: Scripts and logs detailing the training progress and agent performance.
- Clone the repository:
git clone https://github.com/Mrudula-itsjuzme/tic-tac-toe-double-dqn.git
- Install dependencies:
pip install -r requirements.txt
Execute the main entry script to begin the training process:
python main.py- src/: Core implementation of the Double DQN agent and game logic.
- docs/: Architectural notes and research findings.
- FAI_proj_S3/: Directory containing project-specific assets and configurations.
- Integration with multi-agent reinforcement learning (MARL).
- Support for larger board sizes and complex game variants.
- Implementation of a user interface for human-versus-AI gameplay.