A new method for multi-class support vector machines by training least number of classifiers

Ran WANG; Sam KWONG; De-Gang CHEN

doi:10.1109/ICMLC.2011.6016830

A new method for multi-class support vector machines by training least number of classifiers

Ran WANG, Sam KWONG, De-Gang CHEN

Research output: Book Chapters | Papers in Conference Proceedings › Conference paper (refereed) › Research › peer-review

5 Citations (Scopus)

Abstract

How to well apply Support Vector Machine (SVM) technique to multi-class classification problem is an important topic in the area of machine learning. In this paper, we propose a novel method which is different from all the existing ones. By constructing the least number of classifiers, it makes better use of the feature space partition, and can fully eliminate the unclassifiable region. The method is specially designed for 2k-class problems first and could be possibly extended further. We compare the proposed method with several existing ones as one-against-rest (OAR), one-against-one (OAO), decision directed acyclic graph (DDAG), and decision tree (DT) based architecture. Experimental results exhibit good feasibility of the proposed model in term of generalization capability, training time and testing time. © 2011 IEEE.

Original language	English
Title of host publication	Proceedings - International Conference on Machine Learning and Cybernetics
Pages	648-653
DOIs	https://doi.org/10.1109/ICMLC.2011.6016830
Publication status	Published - 2011
Externally published	Yes

Keywords

Hyper-plane
Multi-class classification
Support vector machine
Unclassifiable region

Access to Document

10.1109/ICMLC.2011.6016830

Cite this

@inproceedings{900ee7652b604e5fad79d89dd2168324,

title = "A new method for multi-class support vector machines by training least number of classifiers",

abstract = "How to well apply Support Vector Machine (SVM) technique to multi-class classification problem is an important topic in the area of machine learning. In this paper, we propose a novel method which is different from all the existing ones. By constructing the least number of classifiers, it makes better use of the feature space partition, and can fully eliminate the unclassifiable region. The method is specially designed for 2k-class problems first and could be possibly extended further. We compare the proposed method with several existing ones as one-against-rest (OAR), one-against-one (OAO), decision directed acyclic graph (DDAG), and decision tree (DT) based architecture. Experimental results exhibit good feasibility of the proposed model in term of generalization capability, training time and testing time. {\textcopyright} 2011 IEEE.",

keywords = "Hyper-plane, Multi-class classification, Support vector machine, Unclassifiable region",

author = "Ran WANG and Sam KWONG and De-Gang CHEN",

year = "2011",

doi = "10.1109/ICMLC.2011.6016830",

language = "English",

pages = "648--653",

booktitle = "Proceedings - International Conference on Machine Learning and Cybernetics",

}

TY - GEN

T1 - A new method for multi-class support vector machines by training least number of classifiers

AU - WANG, Ran

AU - KWONG, Sam

AU - CHEN, De-Gang

PY - 2011

Y1 - 2011

N2 - How to well apply Support Vector Machine (SVM) technique to multi-class classification problem is an important topic in the area of machine learning. In this paper, we propose a novel method which is different from all the existing ones. By constructing the least number of classifiers, it makes better use of the feature space partition, and can fully eliminate the unclassifiable region. The method is specially designed for 2k-class problems first and could be possibly extended further. We compare the proposed method with several existing ones as one-against-rest (OAR), one-against-one (OAO), decision directed acyclic graph (DDAG), and decision tree (DT) based architecture. Experimental results exhibit good feasibility of the proposed model in term of generalization capability, training time and testing time. © 2011 IEEE.

AB - How to well apply Support Vector Machine (SVM) technique to multi-class classification problem is an important topic in the area of machine learning. In this paper, we propose a novel method which is different from all the existing ones. By constructing the least number of classifiers, it makes better use of the feature space partition, and can fully eliminate the unclassifiable region. The method is specially designed for 2k-class problems first and could be possibly extended further. We compare the proposed method with several existing ones as one-against-rest (OAR), one-against-one (OAO), decision directed acyclic graph (DDAG), and decision tree (DT) based architecture. Experimental results exhibit good feasibility of the proposed model in term of generalization capability, training time and testing time. © 2011 IEEE.

KW - Hyper-plane

KW - Multi-class classification

KW - Support vector machine

KW - Unclassifiable region

UR - http://www.scopus.com/inward/record.url?scp=80155134782&partnerID=8YFLogxK

U2 - 10.1109/ICMLC.2011.6016830

DO - 10.1109/ICMLC.2011.6016830

M3 - Conference paper (refereed)

SP - 648

EP - 653

BT - Proceedings - International Conference on Machine Learning and Cybernetics

ER -

A new method for multi-class support vector machines by training least number of classifiers

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this