The Fast Computation Methods for Extreme Learning Machine
Tao Dou, Xu Zhou · 2015
The extreme learning machine (ELM) that is proposed by Huang is designed based on single-hidden layer feedforward neural networks (SLFNs), which can randomly choose the parameters of hidden nodes and the output weights gotten analytically.So it can get the solution fastly.However, the learning time of ELM is mainly spent on calculating the Moore-Penrose generalized inverse matrices of the hidden layer output matrix.This paper mainly focuses on the effective computation of the Moore-Penrose generalized inverse matrices for ELM.Moreover, several methods are proposed, which are tensor product matrix ELM (TPM-ELM), Geninv ELM Numerical experiments show that both Geninv-ELM and TPM-ELM are faster than other kinds of ELM and can reach comparable generalization performance.