Dual neural gas based structure ensemble with the bagging technique

Author

Chen, Hantao ; Zhang, Xiaodong ; You, Jane ; Han, Guoqiang ; Li, Le

Author_Institution

Sch. of Comput. Sci. & Eng., South China Univ. of Technol., Guangzhou, China

Volume

4

fYear

2012

fDate

15-17 July 2012

Firstpage

1400

Lastpage

1405

Abstract

It is widely accepted that cluster ensemble can improve accuracy, stableness and robustness when compared with single cluster approach. As the bagging technique can enhance the prediction accuracy of unstable learning algorithms, and the neural gas algorithm can achieve the structure of datasets, we propose a new structure ensemble framework, named as dual neural gas based structure ensemble with the bagging technique. Experiments on both UCI datasets and synthetic datasets show that tne new framework works well.

Keywords

learning (artificial intelligence); neural nets; pattern clustering; sampling methods; set theory; VCI datasets; bagging technique; dual neural gas-based structure ensemble; neural gas algorithm; prediction accuracy; resampling technique; robustness; stableness; synthetic datasets; unstable learning algorithm; Abstracts; Bagging; Neural gas; Normalized mutual information; Purity; Structure ensemble;

fLanguage

English

Publisher

ieee

Conference_Titel

Machine Learning and Cybernetics (ICMLC), 2012 International Conference on

Conference_Location

Xian

ISSN

2160-133X

Print_ISBN

978-1-4673-1484-8

Type

conf

DOI

10.1109/ICMLC.2012.6359570

Filename

6359570