Reduced training for hierarchical incremental class learning

Chunyu Bao; Sheng Uei Guan

doi:10.1109/ICCIS.2006.252321

Reduced training for hierarchical incremental class learning

Chunyu Bao^*, Sheng Uei Guan

^*Corresponding author for this work

National University of Singapore

Research output: Chapter in Book or Report/Conference proceeding › Conference Proceeding › peer-review

1 Citation (Scopus)

Abstract

Hierarchical Incremental Class Learning (HICL), proposed by Guan and Li in 2002 [13], 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.

Original language	English
Title of host publication	2006 IEEE Conference on Cybernetics and Intelligent Systems
DOIs	https://doi.org/10.1109/ICCIS.2006.252321
Publication status	Published - 2006
Externally published	Yes
Event	2006 IEEE Conference on Cybernetics and Intelligent Systems - Bangkok, Thailand Duration: 7 Jun 2006 → 9 Jun 2006

Publication series

Name	2006 IEEE Conference on Cybernetics and Intelligent Systems

Conference

Conference	2006 IEEE Conference on Cybernetics and Intelligent Systems
Country/Territory	Thailand
City	Bangkok
Period	7/06/06 → 9/06/06

Keywords

Classifier systems
HICL
Hierarchical learning
Reduced training set

Access to Document

10.1109/ICCIS.2006.252321

Cite this

@inproceedings{1dec72e767d54a9ba336a3bba00ac3da,

title = "Reduced training for hierarchical incremental class learning",

abstract = "Hierarchical Incremental Class Learning (HICL), proposed by Guan and Li in 2002 [13], 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 = "Classifier systems, HICL, Hierarchical learning, Reduced training set",

author = "Chunyu Bao and Guan, {Sheng Uei}",

year = "2006",

doi = "10.1109/ICCIS.2006.252321",

language = "English",

isbn = "1424400236",

series = "2006 IEEE Conference on Cybernetics and Intelligent Systems",

booktitle = "2006 IEEE Conference on Cybernetics and Intelligent Systems",

note = "2006 IEEE Conference on Cybernetics and Intelligent Systems ; Conference date: 07-06-2006 Through 09-06-2006",

}

TY - GEN

T1 - Reduced training for hierarchical incremental class learning

AU - Bao, Chunyu

AU - Guan, Sheng Uei

PY - 2006

Y1 - 2006

N2 - Hierarchical Incremental Class Learning (HICL), proposed by Guan and Li in 2002 [13], 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.

AB - Hierarchical Incremental Class Learning (HICL), proposed by Guan and Li in 2002 [13], 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.

KW - Classifier systems

KW - HICL

KW - Hierarchical learning

KW - Reduced training set

UR - http://www.scopus.com/inward/record.url?scp=84894171754&partnerID=8YFLogxK

U2 - 10.1109/ICCIS.2006.252321

DO - 10.1109/ICCIS.2006.252321

M3 - Conference Proceeding

AN - SCOPUS:84894171754

SN - 1424400236

SN - 9781424400232

T3 - 2006 IEEE Conference on Cybernetics and Intelligent Systems

BT - 2006 IEEE Conference on Cybernetics and Intelligent Systems

T2 - 2006 IEEE Conference on Cybernetics and Intelligent Systems

Y2 - 7 June 2006 through 9 June 2006

ER -

Reduced training for hierarchical incremental class learning

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Cite this