Minesweeper AI

What is Mulago?

Multi-Square or better known as “Mulago” is a competitive game of “beating the machine”. The AI was named after my dear friend “Mulago”, who in fact has memorised the entire biology textbook word for word. The Multi-Square AI on the other hand does not need to know all the possible moves. As the game progresses the AI only needs calculate the most likely moves based on previous successful moves.

A set of rules are declared at the beginning to ensure Mulago sticks to the rules of the game. When the game starts the AI has ability to monitor the competitors progress and moves made. This is fair as the user can also see the moves the AI has made.

Objective?

The objective of Mulago is for the user to try and beat the game, behind each user’s entry the algorithm incrementally calculates possible outcomes. Thus, the program is continually improving and gaining a competitive advantage. Multi-Square increases its own current predicted success rate and improves on its next predetermined moves. At the end of the game the AI will have recorded all the successful moves made by the player and the AI.

How does it work?

At the start of the player and AI are given the same minesweeper level. 5 minutes is set on the clock and the round begins. The first move is an instance of luck and both the AI and the user needs to make a random move. Pseudo random is used for this purpose, but in a later version the AI will be equipped with a truer random function. An example would be using the number of bubbles in a lava lamp as a source of true random.

The AI may take longer to calculate certain areas on the board, which could cost the game. The human player on the other hand can see the level as a whole and if there are any easy moves to be made the human will do those first. The AI may learn to wait for the user to make these easy moves as more rounds are played.