The Selection and Reduction of Fuzzy Rules Based on the Least Angle Regression Algorithm

The rule exploration is a significant bottleneck which restricts the development of the fuzzy system identification. Many rules selection algorithms have been suggested to overcome this issue. In this paper, we also propose a rules selection-reduction algorithm based on least angle regression algorithm. Through sparse representation of fuzzy consequent parameters, the essential rules are selected and the system is simplified. The least angle regression (LARS) solution method for spare representation make the solving process completed in several steps. The experiment results prove the algorithm has excellent performance.