Bayesian Regularized Committee of Extreme Learning Machine

José M. Martínez-Martínez, Pablo Escandell-Montero, Emilio Soria-Olivas, Joan Vila-Francés, Rafael Magdalena-Benedito

2013

Abstract

Extreme Learning Machine (ELM) is an efficient learning algorithm for Single-Hidden Layer Feedforward Networks (SLFNs). Its main advantage is its computational speed due to a random initialization of the parameters of the hidden layer, and the subsequent use of Moore-Penrose’s generalized inverse in order to compute the weights of the output layer. The main inconvenient of this technique is that as some parameters are randomly assigned and remain unchanged during the training process, they can be non-optimum and the network performance may be degraded. This paper aims to reduce this problem using ELM committees. The way to combine them is to use a Bayesian linear regression due to its advantages over other approaches. Simulations on different data sets have demonstrated that this algorithm generally outperforms the original ELM algorithm.

Paper Citation

in Harvard Style

M. Martínez-Martínez J., Escandell-Montero P., Soria-Olivas E., Vila-Francés J. and Magdalena-Benedito R. (2013). Bayesian Regularized Committee of Extreme Learning Machine . In Proceedings of the 2nd International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM, ISBN 978-989-8565-41-9, pages 109-114. DOI: 10.5220/0004173901090114

in Bibtex Style

@conference{icpram13,
author={José M. Martínez-Martínez and Pablo Escandell-Montero and Emilio Soria-Olivas and Joan Vila-Francés and Rafael Magdalena-Benedito},
title={Bayesian Regularized Committee of Extreme Learning Machine},
booktitle={Proceedings of the 2nd International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM,},
year={2013},
pages={109-114},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0004173901090114},
isbn={978-989-8565-41-9},
}

in EndNote Style

TY - CONF
JO - Proceedings of the 2nd International Conference on Pattern Recognition Applications and Methods - Volume 1: ICPRAM,
TI - Bayesian Regularized Committee of Extreme Learning Machine
SN - 978-989-8565-41-9
AU - M. Martínez-Martínez J.
AU - Escandell-Montero P.
AU - Soria-Olivas E.
AU - Vila-Francés J.
AU - Magdalena-Benedito R.
PY - 2013
SP - 109
EP - 114
DO - 10.5220/0004173901090114