Robust TSK Fuzzy Modeling with Proper Clustering Structure
Chih-Ching Hsiao, Shun‐Feng Su · 2005
Abstract: Traditional approaches for modeling TSK fuzzy rules are trying to adjust the parameters in models, and not considering the training data distribution. Hence it will result in an improper clustering structure, especially, when outliers exist. In this paper, a clustering algorithm termed as Robust Proper Structure Fuzzy Regression Algorithm (RPSFR) is proposed to define fuzzy subspaces in a fuzzy regression manner and also data clustering with robust capability against outliers.