Fuzzy Rules Reduction Based on the Least Angle Regression Algorithm
Huiqin Jiang, Xiang Zhang · 2022
High-dimensional data bring ‘Rule Explosion’. How to reduce the fuzzy rules becomes a novel research hotspot. In this paper, we proposed a rules-reduction model. In this proposed model we employ the idea of the sparse reconstruction algorithm to reduce the redundant rule, and transform the fuzzy rule reduction problem into the optimum solution of sparse reconstruction algorithms. We first compute fuzzy basis functions in order to represent the output signal. The expression function of the outputs signal can be seen as the sparse representation function with unknown parameters. Least angle regression algorithm is employed as an optimal method to solve the sparse reconstruction model. In the last section we use the proposed model to model time serial data, our algorithm performs better than Genfis2 and Genfis3.