Differential evolution based on-line feature analysis in an asymmetric subsethood product fuzzy neural network

This paper proposes a novel differential evolution learning based online feature selection method in an asymmetric subsethood product fuzzy neural network (ASuPFuNIS). The fuzzy neural network has fuzzy weights modeled by asymmetric Gaussian fuzzy sets, mutual subsethood based activation spread, product aggregation operator that works in conjunction with volume defuzzification in a differential evolution learning framework. By virtue of a mixed floating point-binary genetic coding and a customized dissimilarity based bit flipping operator, the differential evolution based asymmetric subsethood product network is shown to have online feature selection capabilities on a synthetic data set.