An Improved Fuzzy Extreme Learning Machine for Classification and Regression
Chen‐Sen Ouyang, Tzu-Chin Kao, Yu-Yuan Cheng, Chih‐Hung Wu, Chiung-Hui Tsai, Mengwei Wu · 2016
Wong et al. proposed a fuzzy extreme learning machine (F-ELM) which possessed advantages of fuzzy inference systems and extreme learning machines. However, the generalization capability and flexibility of F-ELM are restricted by constant rule consequences and the generalized AND operator. Therefore, first-order Takagi-Sugeno-Kang (TSK) type fuzzy rule consequences and a compensatory fuzzy operator are introduced to replace original ones for enhancing the generalization capability and flexibility of F-ELM. Compared with the F-ELM, experimental results have shown the improved F-ELM produces the higher classification accuracy for classification problems and the lower mean squared errors for regression problems, and possesses the better stability.