An SVD-based fuzzy model reduction strategy

A numerically reliable orthogonal transformation technique, known as the singular value decomposition (SVD), is utilized to detect and select the dominant fuzzy rules from a rule base using a nonlinear limit cycle modeling problem.

Abstract

This paper describes a novel fuzzy model reduction approach for overcoming the curse of dimensionality associated with high-dimensional data modeling problems. A numerically reliable orthogonal transformation technique, known as the singular value decomposition (SVD), is utilized to detect and select the dominant fuzzy rules from a rule base. The effectiveness of the proposed approach is illustrated using a nonlinear limit cycle modeling problem.