A novel reformulated radial basis function neural network

Abstract

Single-hidden-layer feedforward networks (SLFNs) with radial basis function (RBF) hidden nodes are universal approximators when all the parameters of the networks are allowed adjustable. The learning speed of SLFNs is in general far slower than required and it has been a major bottleneck in their applications for past decades) Huang et al. propose a new learning algorithm called extreme learning machine (ELM) for SLFNs which randomly chooses hidden nodes and analytically determines the output weights. In this paper, common choices of RBF for generating ELM are analyzed and compared. The purpose of this study is to explore comparative strengths and weaknesses of the choices and to show some useful guidelines on how to choose an appropriate RBF hidden nodes for a particular problem.