Abstract:
Much current research in the game AI community has been more explored and better understood. It is interesting to quest what this could mean for games as we stumble towards human-level intelligence in different kind of applications. The popular Puzzle Games, Light Up and Sudoku, provided the wide leisure market in magazines and application software. Light Up is a single-player puzzle game from Nikoli in Japan. The game rules are simple, easy to get started but difficult to master. The design and implementation of logic strategy for solving Light Up is contributed to promote the inference ability for players. The main purposes of this paper are to conduct the pattern matching concept to improve the drawback of elimination search, and to propose the hybrid Multi-Patterns with Monte-Carlo Search (MPMCS) approach to enhance the performance of the searching ability for the Light Up problem. Several Light Up problems with various sizes are considered to compare the proposed MPMCS algorithm. Experimental results show that the MPMCS is efficiently to solve for the Light Up. In addition, the development of MPMCS approach can be guided in the dynamic simulation and can be useful for auxiliary education purposes.
|