Fuzzy search strategy generation for adversarial systems using fuzzy process particle swarm optimization, fuzzy patterns, and a hunch factor.

摘要

Adversarial game-playing situations have been studied in both game theory and artificial intelligence since the 1950s. However, developing strategies for game playing is challenging due to the large search tree size as the environment size increases. The research presented develops a centralized fuzzy search strategy to find winning strategies for adversarial situations using fuzzy logic. The aforementioned fuzzy algorithm, Fuzzy Strategy Finder (FSF), can be used to find valid strategies for any problem that can be written as a Reti-like or zero-sum game. Using several chess endgames, the FSF will be shown to have a faster runtime without a significant negative impact on the quality of solutions. The FSF contains three key components: Fuzzy Pattern Generation (FPG), Fuzzy Process Particle Swarm Optimization (FP2SO), and the hunch factor.;The Fuzzy Pattern Generation (FPG) algorithm provides an essential tool to break a large environment into smaller interest areas to analyze and select possible moves. FPG improves upon the existing zone concept from Stilman's Linguistic Geometry algorithm in two ways: quicker to compute and producing significantly smaller number of patterns. The FPG portion of the FSF algorithm is shown to narrow the overall search space for FSF by up to 80%. Additionally, it was proven that FPG does not increase the search space size.;The addition of fuzzy logic to the Particle Swarm Optimization (PSO) strategy introduces a level of abstraction. The FP2SO algorithm introduces fuzziness to the algorithm on two levels: the data and the process. Considering individual elements of data as a membership function fuzzifies the data. Replacing the traditional operators with fuzzy equivalent operators fuzzifies the PSO process. The benefit of introducing fuzzy logic to PSO is to allow the system to encompass a more human-like decision process and to increase runtime. The FP2SO is shown, when calibrated properly, to run faster than the non-fuzzy PSO with minimal impact to accuracy.;A third component the hunch factor supplies a human "hunch-like" element into the decision-making processes of the FSF algorithm. Additionally, the hunch acts a fuzzy learning component for FSF since the hunch is continually altered during FSF execution. The hunch is shown to increase the accuracy of FSF.