Neural Network Learner for Minesweeper
This is an incremental improvement for Minesweeper solvers, offering a novel machine learning approach to a logic-based game.
The authors tackled the NP-hard problem of solving Minesweeper by developing a neural network-based learner, which competes well with CSP solvers in Beginner and Intermediate modes, achieving high success rates but with slower performance than deterministic solvers.
Minesweeper is an interesting single player game based on logic, memory and guessing. Solving Minesweeper has been shown to be an NP-hard task. Deterministic solvers are the best known approach for solving Minesweeper. This project proposes a neural network based learner for solving Minesweeper. To choose the best learner, different architectures and configurations of neural networks were trained on hundreds of thousands of games. Surprisingly, the proposed neural network based learner has shown to be a very good approximation function for solving Minesweeper. The neural network learner competes well with the CSP solvers, especially in Beginner and Intermediate modes of the game. It was also observed that despite having high success rates, the best neural learner was considerably slower than the best deterministic solver. This report also discusses the overheads and limitations faced while creating highly successful neural networks for Minesweeper.