N-ary decomposition for multi-class classification

Joey Tianyi Zhou; Ivor W. Tsang; Shen Shyang Ho; Klaus Robert Müller

doi:10.1007/s10994-019-05786-2

N-ary decomposition for multi-class classification

Joey Tianyi Zhou
, Ivor W. Tsang
, Shen Shyang Ho
, Klaus Robert Müller

Computer Science

Rowan University

Research output: Contribution to journal › Article › peer-review

8 Scopus citations

Abstract

A common way of solving a multi-class classification problem is to decompose it into a collection of simpler two-class problems. One major disadvantage is that with such a binary decomposition scheme it may be difficult to represent subtle between-class differences in many-class classification problems due to limited choices of binary-value partitions. To overcome this challenge, we propose a new decomposition method called N-ary decomposition that decomposes the original multi-class problem into a set of simpler multi-class subproblems. We theoretically show that the proposed N-ary decomposition could be unified into the framework of error correcting output codes and give the generalization error bound of an N-ary decomposition for multi-class classification. Extensive experimental results demonstrate the state-of-the-art performance of our approach.

Original language	English (US)
Pages (from-to)	809-830
Number of pages	22
Journal	Machine Learning
Volume	108
Issue number	5
DOIs	https://doi.org/10.1007/s10994-019-05786-2
State	Published - May 15 2019

ASJC Scopus subject areas

Software
Artificial Intelligence

Access to Document

10.1007/s10994-019-05786-2

Cite this

@article{68eccd3c32624df58fe932200ac8539f,

title = "N-ary decomposition for multi-class classification",

abstract = "A common way of solving a multi-class classification problem is to decompose it into a collection of simpler two-class problems. One major disadvantage is that with such a binary decomposition scheme it may be difficult to represent subtle between-class differences in many-class classification problems due to limited choices of binary-value partitions. To overcome this challenge, we propose a new decomposition method called N-ary decomposition that decomposes the original multi-class problem into a set of simpler multi-class subproblems. We theoretically show that the proposed N-ary decomposition could be unified into the framework of error correcting output codes and give the generalization error bound of an N-ary decomposition for multi-class classification. Extensive experimental results demonstrate the state-of-the-art performance of our approach.",

author = "Zhou, \{Joey Tianyi\} and Tsang, \{Ivor W.\} and Ho, \{Shen Shyang\} and M{\"u}ller, \{Klaus Robert\}",

note = "Publisher Copyright: {\textcopyright} 2019, The Author(s), under exclusive licence to Springer Science+Business Media LLC, part of Springer Nature.",

year = "2019",

month = may,

day = "15",

doi = "10.1007/s10994-019-05786-2",

language = "English (US)",

volume = "108",

pages = "809--830",

journal = "Machine Learning",

issn = "0885-6125",

publisher = "Springer Netherlands",

number = "5",

}

TY - JOUR

T1 - N-ary decomposition for multi-class classification

AU - Zhou, Joey Tianyi

AU - Tsang, Ivor W.

AU - Ho, Shen Shyang

AU - Müller, Klaus Robert

PY - 2019/5/15

Y1 - 2019/5/15

N2 - A common way of solving a multi-class classification problem is to decompose it into a collection of simpler two-class problems. One major disadvantage is that with such a binary decomposition scheme it may be difficult to represent subtle between-class differences in many-class classification problems due to limited choices of binary-value partitions. To overcome this challenge, we propose a new decomposition method called N-ary decomposition that decomposes the original multi-class problem into a set of simpler multi-class subproblems. We theoretically show that the proposed N-ary decomposition could be unified into the framework of error correcting output codes and give the generalization error bound of an N-ary decomposition for multi-class classification. Extensive experimental results demonstrate the state-of-the-art performance of our approach.

AB - A common way of solving a multi-class classification problem is to decompose it into a collection of simpler two-class problems. One major disadvantage is that with such a binary decomposition scheme it may be difficult to represent subtle between-class differences in many-class classification problems due to limited choices of binary-value partitions. To overcome this challenge, we propose a new decomposition method called N-ary decomposition that decomposes the original multi-class problem into a set of simpler multi-class subproblems. We theoretically show that the proposed N-ary decomposition could be unified into the framework of error correcting output codes and give the generalization error bound of an N-ary decomposition for multi-class classification. Extensive experimental results demonstrate the state-of-the-art performance of our approach.

UR - https://www.scopus.com/pages/publications/85066631663

UR - https://www.scopus.com/pages/publications/85066631663#tab=citedBy

U2 - 10.1007/s10994-019-05786-2

DO - 10.1007/s10994-019-05786-2

M3 - Article

SN - 0885-6125

VL - 108

SP - 809

EP - 830

JO - Machine Learning

JF - Machine Learning

IS - 5

ER -

N-ary decomposition for multi-class classification

Abstract

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this