Designing interval type-2 fuzzy logic systems using an SVD-QR method: Rule reduction
Qilian Liang, Jerry M. Mendel · International Journal of Intelligent Systems · 2000
A type-2 fuzzy logic system (FLS) can handle numerical and linguistic uncertainties, but, like a type-1 FLS, rule explosion is one of its major disadvantages. In this paper, we present a design method which can tremendously reduce rule number for interval type-2 fuzzy logic systems using an SVD-QR method. The SVD-QR method is performed after extracting two fuzzy basis function expansions from the interval type-2 FLS. We evaluate this method by applying it to a time-series forecasting problem in conjunction with back-propagation training, and demonstrate that tremendous rule number reduction ratio is achieved with very little performance degradation. © 2000 John Wiley & Sons, Inc.