A comparison of consensus- and critical point-based classification strategies

Author

Weichao Xu ; Rubao Ma ; Qinruo Wang

Author_Institution

Sch. of Autom., Guangdong Univ. of Technol., Guangzhou, China

fYear

2012

fDate

12-14 July 2012

Firstpage

55

Lastpage

58

Abstract

In this paper we compare the consensus-based strategy (CBS) and the critical point-based strategy (CPBS) which are commonly adopted in the practice of designing classifiers. Theoretical analyses and simulation results reveal the close relationship between the kurtosis (long tailedness) of the distribution of data patterns and the performance of SVM designed with CPBS. Monte Carlo simulation results agree with the theoretical findings.

Keywords

Monte Carlo methods; pattern classification; support vector machines; CBS; CPBS; Monte Carlo simulation; SVM; classifier design; consensus-based classification strategy; critical point-based classification strategy; data pattern distribution; kurtosis; support vector machines; Educational institutions; Gaussian distribution; Monte Carlo methods; Random variables; Simulation; Support vector machines; Training; Bernoulli distribution (BD); Consensus-based strategy (CBS); Critical point-based strategy (CPBS); Fisher linear discrimination analysis (FLDA); Laplace distribution (LD); Normal distribution (ND); Support vector machine (SVM); Uniform distribution (UD);

fLanguage

English

Publisher

ieee

Conference_Titel

Computational Intelligence and Cybernetics (CyberneticsCom), 2012 IEEE International Conference on

Conference_Location

Bali

Print_ISBN

978-1-4673-0891-5

Type

conf

DOI

10.1109/CyberneticsCom.2012.6381616

Filename

6381616