TSK 퍼지 규칙과 랜덤 클러스터링을 이용한 ELM 예측기의 설계

Abstract

Yeom, Chan-Uk Advisor : Prof. Kwak, Keun Chang, Ph. D. Dept. of Control and Instrumentation Eng., Graduate School of Chosun University In this paper propose an ELM predictor using TSK fuzzy rule and random clustering method to improve the prediction performance of ELM predictor. The TSK-based ELM model consists of a structure that uses a linear function in place of a weight and sets the center of the cluster at random. There is no weight between the input layer and the hidden layer, and the weight between the hidden layer and the output layer is a linear equation. We propose method generates meaningful rules by using the if-then rule. Also, it’s improves the performance of the predictor by randomly generating the cluster center through the initial membership matrix. Experiments were compared with existing ELM predictors using function approximation, Boston Housing, and Auto-MPG data. Experimental results showed that the proposed method showed better performance than the conventional ELM predictor.