Monte Carlo tree search in Einstein Wurfelt Niccht!

Includes bibliographical references (page 26)Monte Carlo simulations have been often used for artificial intelligence with many games. When combined with a search tree and millions of random simulations per second, it can produce a recommended move according to the simulation strategy. But, results can be different for each type of game. The structure and search formula for the tree must be tailored differently for each type of game. In this project, I have created a program called SteinBot for playing Einstein Wurfelt Nicht, which used an improved version of the MCTS algorithm to produce a recommended move. The MCTS was created in a special way to support die values. SteinBot was tested with hundreds of games with the random method and confirmed that improvements are needed for the random simulations and search formula. SteinBot played against many humans with the improved strategy and won 63% of the games