Inspiration: After making a minimax engine for connect 4 the computational load on my computer was immense so we thought of training a neural network to play connect 4 in a less computationally intensive manner.

What it does: Plays connect 4 against a human opponent using pygame.

How we built it: We built a minimax engine for connect 4 and then used that to generate a training data set for a classification neural network.

Challenges we ran into: Time was a big issue as we were only able to train the network for 1 hour.

Accomplishments that we're proud of: Creating a neural network from the ground up without using any external libraries/apis

What we learned: Training a neural network take a long time.

What's next for Connect 4 Neural Network: Implementing a Q learning method so that the AI can play itself over and over again and learn from it as a result.

Built With

Share this project:

Updates