Novel Self-Organizing Takagi Sugeno Kang Fuzzy Neural Networks Based on ART-like Clustering

The existing Self-Organizing Takagi Sugeno Kang Fuzzy Neural Networks (S-TSKfnn) structure uses virus infection clustering (VIC) method to generate fuzzy rules. In this paper, we propose a novel architecture called Modified S-TSKfnn (MS-TSKfnn) that uses ART-like clustering called discrete incremental clustering (DIC). By doing so, MS-TSKfnn is able to handle online data input, and its performance is also enhanced. Most importantly, the accurate clustering in the fuzzy set derivation has significantly reduced the number of fuzzy TSK rules necessary to describe a problem. Extensive simulations are conducted using MS-TSKfnn and its performance is encouraging when benchmarked against other established neuro-fuzzy systems. The empirical work also firmly demonstrated the importance of clustering within a fuzzy neural reasoning system in ensuring a compact and expressive fuzzy rate base.