I have been working on recurrent algorithms. They’re a bunch of examples in which you can implement these algorithms. However, we usually see that they are efficient.
Connect 4 is a simple game that can be easily programmed and solved to win every time. I have made in Python an AI that solves and wins. It is not programmed in C++ because I wanted a GUI.
Then I created a C++ project where I exploited the performance and reach 8 depth in a few seconds.
This is an example of a game I have played against the AI. It performs really well 🙂
Performance
- It performs well, human-level, and sometimes overperforming.
Next steps
- Parallel computing
- Optimize algorithm
- Convolutional neural network vs Recurrent algorithm
- implementation of a complex algorithm
- different search algorithms (Min-max searching, alpha-beta pruning, null move heuristic, quiescence searching, etc), better than brute force search
Repository
GitHub repository: C++ repository