Move Ordering Using Neural Networks
The efficiency of alpha-beta search algorithms heavily de- pends on the order in which the moves are examined. This paper focuses on using neural networks to estimate the likelihood of a move being the best in a certain position. The moves considered more likely to be the best are examined first. We selected Lines of Action as a testing ground. We investigated several schemes to encode the moves in a neural net- work. In the experiments, the best performance was obtained by using one output unit for each possible move of the game. The results indicate that our move-ordering approach can speed up the search with 20 to 50 percent compared with one of the best current alternatives, the history heuristic.
Unable to display preview. Download preview PDF.
- 2.Kocsis, L., Uiterwijk, J. W. H. M., Herik, H. J. van den: Learning Time Allocation using Neural Networks. Working Notes of the Second International Conference on Computers and Games (2000) 297–314Google Scholar
- 3.Riedmiller, M., Braun, H.: A Direct Adaptive Method for Faster Backpropagation Learning: The RPROP Algorithm. Proceedings of the IEEE International Conference on Neural Networks 1993 (ICNN 93) (1993) 586–591Google Scholar
- 4.Schaeffer, J.: The History Heuristic. ICCA Journal 6(3) (1983) 16–19Google Scholar
- 5.Winands, M. H. M.: Analysis and Implementation of Lines of Action. MSc thesis CS 00-03, Department of Computer Science, Universiteit Maastricht (2000)Google Scholar