Rule-by-rule input significance analysis in fuzzy system modeling

Input or feature selection is one the most important steps of system modeling. Elimination of irrelevant variables can save time, money and can improve the precision of model that we are trying to discover. In fuzzy system modeling (FSM) approaches, input selection plays an important role too. The input selection algorithms that are under our investigation did not consider one crucial fact. An input variable may of may not be significant in a specific rule, not in overall system. In this paper, an input selection algorithm that takes this observation into account is proposed as an extension of the input selection algorithms found in the literature. The proposed algorithm is applied on a nonlinear function and successful results are achieved.