Robust incremental growing multi-experts network

Most supervised neural networks are trained by minimizing the mean square error (MSE) of the training set. In the presence of outliers, the resulting neural network model can differ significantly from the underlying model that generates the data. This paper outlines two robust learning methods for a...

Full description

Bibliographic Details
Main Authors: LOO, C, RAJESWARI, M, RAO, M
Format: Article
Published: 2006
Subjects:
Online Access:http://shdl.mmu.edu.my/2057/
_version_ 1848789952601522176
author LOO, C
RAJESWARI, M
RAO, M
author_facet LOO, C
RAJESWARI, M
RAO, M
author_sort LOO, C
building MMU Institutional Repository
collection Online Access
description Most supervised neural networks are trained by minimizing the mean square error (MSE) of the training set. In the presence of outliers, the resulting neural network model can differ significantly from the underlying model that generates the data. This paper outlines two robust learning methods for a dynamic structure neural network called incremental growing multi-experts network (IGMN). It is convincingly shown by simulation that by using a scaled robust objective function instead of the least squares function, the influence of the outliers in the training data can be completely eliminated. The network generates a much better approximation in the neighborhood of outliers. Thus, the two proposed robust learning methods namely robust least mean squares (RLMSs) and least mean log squares (LMLSs) are insensitive to the presence of outliers unlike the least mean squares (LMSs) cost function. Moreover, various types of supervised learning algorithms can easily adopt LMLS, which is a parameter-free method. (C) 2005 Elsevier B. V. All rights reserved.
first_indexed 2025-11-14T18:04:54Z
format Article
id mmu-2057
institution Multimedia University
institution_category Local University
last_indexed 2025-11-14T18:04:54Z
publishDate 2006
recordtype eprints
repository_type Digital Repository
spelling mmu-20572011-08-10T06:39:15Z http://shdl.mmu.edu.my/2057/ Robust incremental growing multi-experts network LOO, C RAJESWARI, M RAO, M QA75.5-76.95 Electronic computers. Computer science Most supervised neural networks are trained by minimizing the mean square error (MSE) of the training set. In the presence of outliers, the resulting neural network model can differ significantly from the underlying model that generates the data. This paper outlines two robust learning methods for a dynamic structure neural network called incremental growing multi-experts network (IGMN). It is convincingly shown by simulation that by using a scaled robust objective function instead of the least squares function, the influence of the outliers in the training data can be completely eliminated. The network generates a much better approximation in the neighborhood of outliers. Thus, the two proposed robust learning methods namely robust least mean squares (RLMSs) and least mean log squares (LMLSs) are insensitive to the presence of outliers unlike the least mean squares (LMSs) cost function. Moreover, various types of supervised learning algorithms can easily adopt LMLS, which is a parameter-free method. (C) 2005 Elsevier B. V. All rights reserved. 2006-01 Article NonPeerReviewed LOO, C and RAJESWARI, M and RAO, M (2006) Robust incremental growing multi-experts network. Applied Soft Computing, 6 (2). pp. 139-153. ISSN 15684946 http://dx.doi.org/10.1016/j.asoc.2004.12.003 doi:10.1016/j.asoc.2004.12.003 doi:10.1016/j.asoc.2004.12.003
spellingShingle QA75.5-76.95 Electronic computers. Computer science
LOO, C
RAJESWARI, M
RAO, M
Robust incremental growing multi-experts network
title Robust incremental growing multi-experts network
title_full Robust incremental growing multi-experts network
title_fullStr Robust incremental growing multi-experts network
title_full_unstemmed Robust incremental growing multi-experts network
title_short Robust incremental growing multi-experts network
title_sort robust incremental growing multi-experts network
topic QA75.5-76.95 Electronic computers. Computer science
url http://shdl.mmu.edu.my/2057/
http://shdl.mmu.edu.my/2057/
http://shdl.mmu.edu.my/2057/