Reduced Training for Hierarchical Incremental Class Learning

Author

Bao, Chunyu ; Guan, Sheng-Uei

Author_Institution

Dept. of Electr. & Comput. Eng., National Univ. of Singapore

fYear

2006

fDate

7-9 June 2006

Firstpage

1

Lastpage

6

Abstract

Hierarchical incremental class learning (HICL), proposed by Guan and Li in 2002, is a recently proposed task decomposition method that addresses the pattern classification problem. HICL is proven to be a good classifier but closer examination reveals areas for potential improvement. This paper presents an approach to improve the classification accuracy of HICL by applying the concept of reduced pattern training (RPT). The procedure for RPT is described and compared with the original training procedure. RPT systematically reduces the size of the training data set based on the order of sub-networks built. The results from benchmark classification problems show much promise for the improved model

Keywords

learning (artificial intelligence); pattern classification; classifier system; hierarchical incremental class learning; pattern classification; reduced pattern training; reduced training set; task decomposition; Biological neural networks; Computer networks; Crosstalk; Decision making; Interference; Jacobian matrices; Neural networks; Parallel processing; Pattern classification; Training data; HICL; classifier systems; hierarchical learning; reduced training set;

fLanguage

English

Publisher

ieee

Conference_Titel

Cybernetics and Intelligent Systems, 2006 IEEE Conference on

Conference_Location

Bangkok

Print_ISBN

1-4244-0023-6

Type

conf

DOI

10.1109/ICCIS.2006.252321

Filename

4017880