Rule Selection in Fuzzy Systems using Heuristics and Branch Prediction

Rule bases, providing complete information about the system at hand in a fuzzy logic controller, tend to be huge. Selecting rules that should be applied to the current inputs of the system becomes an increasingly complex task, as the rule base size increases. Techniques have been developed to provide the relevant rules for inference to improve the speed of operation of fuzzy systems. This paper proposes an approach using a simple heuristic to identify a most probable set of rules and then predict the rule that will be used for the current system inputs. A prediction strategy, used for branch prediction in processors, is employed in predicting the rule to be used. The proposed approach has been compared with few approaches for rule selection and results are provided