Generalized extreme learning machine acting on a metric space

Generalized Extreme Learning Machine (GELM) is a kind of fast and efficient learning algorithm for training Generalized Single-layer hidden Feedforward Networks (GSLFNs) acting on some metric spaces. However, noisy data often produce over-fitting phenomena in practical applications. Therefore, an improved learning algorithm, called Regularized Generalized Extreme Learning Machine (R-GELM), is proposed with regularization method to improve the generalization of GELM. Experimental comparisons of the proposed R-GELM are carried out with four state-of-the-art algorithms, and the experimental results show that the proposed R-GELM has better generalization and less computational cost than the others.