Toward an optimal kernel extreme learning machine using a chaotic moth-flame optimization strategy with applications in medical diagnoses

Abstract

This study proposes a novel learning scheme for the kernel extreme learning machine (KELM) based on the chaotic moth-flame optimization (CMFO) strategy. In the proposed scheme, CMFO simultaneously performs parameter optimization and feature selection. The proposed methodology is rigorously compared to several other competitive KELM models that are based on the original moth-flame optimization, particle swarm optimization, and genetic algorithms. The comparison is made using the medical diagnosis problems of Parkinson's disease and breast cancer. And the proposed method has successfully been applied to practical medical diagnosis cases. The experimental results demonstrate that, compared to the alternative methods, the proposed method offers significantly better classification performance and also obtains a smaller feature subset. Promisingly, the proposed CMFOFS-KELM, can serve as an effective and efficient computer aided tool for medical diagnosis in the field of medical decision making.