Neural Networks in solving Minesweeper

The aim of this project is to show efficiency of chosen neural network implementations in solving Minesweeper when compared to much simpler analytical solutions. This repository contains all the code written for the needs of this project as well as directory of result files used to create documentation.

What files to run and in what order:

1. Run _0_solve_human.py to play the game manually (optional)
2. Run _1_solve_anaytical.py and _1_solve_analytical_5x5.py to create datasets used in training nn models
3. Run _2_train_MODEL.py files to create and train models (adjust SIZE accordingly to dataset you want to use)
4. Run _3_solve_MODEL.py files to check how trained models perform (adjust SIZE accordingly to model you want to test)
5. Run _4_train_decision_tree.py to create decision tree showing model preferences when making decisions (uses logs created in point 4)
6. Run _5_display_decision_tree.py to display decision trees from previous point