Genetic programming for model selection of TSK-fuzzy systems

This paper compares a genetic programming (GP) approach with a greedy parti tion algorithm (LOLIMOT) for structure identification of local linear neuro -fuzzy models. The crisp linear conclusion part of a Takagi-Sugeno-Kang (TS K) fuzzy rule describes the underlying model in the local region specified in the premise. The objective of structure identification is to identify an optimal partition of the input space into Gaussian, axis-orthogonal fuzzy sets. The linear parameters in the rule consequent are then estimated by me ans of a local weighted least-squares algorithm. LOLIMOT is an incremental tree-construction algorithm that partitions the input space by axis-orthogo nal splits, In each iteration it greedily adds the new model that minimizes the classification error. GP performs a global search for the optimal part ition tree and is therefore able to backtrack in case of sub-optimal interm ediate split decisions. We compare the performance of both methods for func tion approximation of a highly nonlinear two-dimensional test function and an engine characteristic map. (C) 2001 Elsevier Science Inc, All rights res erved.