CompletedLast commit Jul 2025
Neural Network Labyrinth
A maze game where a neural network learns to navigate procedurally generated levels.
What I built
- A TensorFlow/Keras network steers the character and learns to find the exit.
- Procedurally generated mazes with adjustable complexity.
- Manual play and AI mode side by side, with training on demand.
- My role
- Author
- Stack
- Python
- TensorFlow
- Keras
- Pygame
- NumPy
How it fits together
Drawn from the public repositories. The outline shows who built each part — credit is part of the fact.
- Maze generatoradjustable complexity — Built by me
- Manual play — Built by me
- GamePygame — Built by me
- Neural networkTensorFlow · Keras — Built by me
- Training mode — Built by me
The same, step by step
- Maze generator (adjustable complexity) to Game (Pygame)new level
- Manual play to Game (Pygame)moves
- Game (Pygame) with Neural network (TensorFlow · Keras), both waysstate · next move
- Training mode to Neural network (TensorFlow · Keras)on demand