Multi-Armed Bandits for Minesweeper: Profiting from Exploration-Exploitation Synergy
This work provides a detailed learning perspective on Minesweeper, offering original insights for the puzzle-solving domain, though it is incremental in applying known algorithms to a specific game.
The researchers tackled the problem of creating autonomous players for Minesweeper using a novel reinforcement learning methodology based on Multi-Armed Bandit algorithms, achieving success especially on smaller game boards like the beginner level.
A popular computer puzzle, the game of Minesweeper requires its human players to have a mix of both luck and strategy to succeed. Analyzing these aspects more formally, in our research we assessed the feasibility of a novel methodology based on Reinforcement Learning as an adequate approach to tackle the problem presented by this game. For this purpose we employed Multi-Armed Bandit algorithms which were carefully adapted in order to enable their use to define autonomous computational players, targeting to make the best use of some game peculiarities. After experimental evaluation, results showed that this approach was indeed successful, especially in smaller game boards, such as the standard beginner level. Despite this fact the main contribution of this work is a detailed examination of Minesweeper from a learning perspective, which led to various original insights which are thoroughly discussed.