Input partitioning to mixture of experts

Author

Tang, Bin ; Heywood, Malcolm I. ; Shepherd, Michael

Author_Institution

Fac. of Comput. Sci., Dalhousie Univ., Halifax, NS, Canada

Volume

1

fYear

2002

fDate

6/24/1905 12:00:00 AM

Firstpage

227

Lastpage

232

Abstract

Given a supervised learning context, the mixture of experts approach uses several neural networks in parallel to provide a modular solution to the overall problem.. Under the mixtures of experts architecture a method for ´designing´ the number of experts and assigning local ´regions´ of the input space to individual experts is investigated. Classification performance and transparency of the scheme is found to be significantly better than that using a standard mixtures of experts

Keywords

learning (artificial intelligence); neural nets; pattern classification; pattern clustering; self-organising feature maps; input partitioning; learning algorithm; mixtures of experts; neural networks; pattern classification; potential function clustering; self-organizing feature map; transparency; Computer science; Data preprocessing; Decision trees; Feeds; Jacobian matrices; Machine learning; Neural networks; Piecewise linear techniques; Principal component analysis; Supervised learning;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks, 2002. IJCNN '02. Proceedings of the 2002 International Joint Conference on

Conference_Location

Honolulu, HI

ISSN

1098-7576

Print_ISBN

0-7803-7278-6

Type

conf

DOI

10.1109/IJCNN.2002.1005474

Filename

1005474