MC3: A multi-class consensus classification framework

Tanmoy Chakraborty; Des Chandhok; V. S. Subrahmanian

doi:10.1007/978-3-319-57454-7_27

MC3: A multi-class consensus classification framework

Tanmoy Chakraborty^*, Des Chandhok, V. S. Subrahmanian

^*Corresponding author for this work

Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

2 Scopus citations

Abstract

In this paper, we propose MC3, an ensemble framework for multi-class classification. MC3 is built on “consensus learning”, a novel learning paradigm where each individual base classifier keeps on improving its classification by exploiting the outcomes obtained from other classifiers until a consensus is reached. Based on this idea, we propose two algorithms, MC3-R and MC3-S that make different trade-offs between quality and runtime. We conduct rigorous experiments comparing MC3-R and MC3-S with 12 baseline classifiers on 13 different datasets. Our algorithms perform as well or better than the best baseline classifier, achieving on average, a 5.56% performance improvement. Moreover, unlike existing baseline algorithms, our algorithms also improve the performance of individual base classifiers up to 10%. (The code is available at https://github.com/MC3-code.).

Original language	English (US)
Title of host publication	Advances in Knowledge Discovery and Data Mining - 21st Pacific-Asia Conference, PAKDD 2017, Proceedings
Editors	Kyuseok Shim, Jae-Gil Lee, Longbing Cao, Xuemin Lin, Jinho Kim, Yang-Sae Moon
Publisher	Springer Verlag
Pages	343-355
Number of pages	13
ISBN (Print)	9783319574530
DOIs	https://doi.org/10.1007/978-3-319-57454-7_27
State	Published - 2017
Event	21st Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2017 - Jeju, Korea, Republic of Duration: May 23 2017 → May 26 2017

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	10234 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	21st Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2017
Country/Territory	Korea, Republic of
City	Jeju
Period	5/23/17 → 5/26/17

Funding

Parts of this work were funded by ONR grants N00014-15-R-BA010, N00014-16-R-BA01, N000141612739 and N000141612918.

Keywords

Consensus
Ensemble learning
Multi-class classification

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-319-57454-7_27

Cite this

Chakraborty, T., Chandhok, D., & Subrahmanian, V. S. (2017). MC3: A multi-class consensus classification framework. In K. Shim, J.-G. Lee, L. Cao, X. Lin, J. Kim, & Y.-S. Moon (Eds.), Advances in Knowledge Discovery and Data Mining - 21st Pacific-Asia Conference, PAKDD 2017, Proceedings (pp. 343-355). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10234 LNAI). Springer Verlag. https://doi.org/10.1007/978-3-319-57454-7_27

Chakraborty, Tanmoy ; Chandhok, Des ; Subrahmanian, V. S. / MC3 : A multi-class consensus classification framework. Advances in Knowledge Discovery and Data Mining - 21st Pacific-Asia Conference, PAKDD 2017, Proceedings. editor / Kyuseok Shim ; Jae-Gil Lee ; Longbing Cao ; Xuemin Lin ; Jinho Kim ; Yang-Sae Moon. Springer Verlag, 2017. pp. 343-355 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{6f069c80507b4bcb9235c880b6fa7eb7,

title = "MC3: A multi-class consensus classification framework",

abstract = "In this paper, we propose MC3, an ensemble framework for multi-class classification. MC3 is built on “consensus learning”, a novel learning paradigm where each individual base classifier keeps on improving its classification by exploiting the outcomes obtained from other classifiers until a consensus is reached. Based on this idea, we propose two algorithms, MC3-R and MC3-S that make different trade-offs between quality and runtime. We conduct rigorous experiments comparing MC3-R and MC3-S with 12 baseline classifiers on 13 different datasets. Our algorithms perform as well or better than the best baseline classifier, achieving on average, a 5.56% performance improvement. Moreover, unlike existing baseline algorithms, our algorithms also improve the performance of individual base classifiers up to 10%. (The code is available at https://github.com/MC3-code.).",

keywords = "Consensus, Ensemble learning, Multi-class classification",

author = "Tanmoy Chakraborty and Des Chandhok and Subrahmanian, {V. S.}",

note = "Funding Information: Parts of this work were funded by ONR grants N00014-15-R-BA010, N00014-16-R-BA01, N000141612739 and N000141612918. Publisher Copyright: {\textcopyright} 2017, Springer International Publishing AG.; 21st Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2017 ; Conference date: 23-05-2017 Through 26-05-2017",

year = "2017",

doi = "10.1007/978-3-319-57454-7_27",

language = "English (US)",

isbn = "9783319574530",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "343--355",

editor = "Kyuseok Shim and Jae-Gil Lee and Longbing Cao and Xuemin Lin and Jinho Kim and Yang-Sae Moon",

booktitle = "Advances in Knowledge Discovery and Data Mining - 21st Pacific-Asia Conference, PAKDD 2017, Proceedings",

}

Chakraborty, T, Chandhok, D & Subrahmanian, VS 2017, MC3: A multi-class consensus classification framework. in K Shim, J-G Lee, L Cao, X Lin, J Kim & Y-S Moon (eds), Advances in Knowledge Discovery and Data Mining - 21st Pacific-Asia Conference, PAKDD 2017, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10234 LNAI, Springer Verlag, pp. 343-355, 21st Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2017, Jeju, Korea, Republic of, 5/23/17. https://doi.org/10.1007/978-3-319-57454-7_27

MC3: A multi-class consensus classification framework. / Chakraborty, Tanmoy; Chandhok, Des; Subrahmanian, V. S.
Advances in Knowledge Discovery and Data Mining - 21st Pacific-Asia Conference, PAKDD 2017, Proceedings. ed. / Kyuseok Shim; Jae-Gil Lee; Longbing Cao; Xuemin Lin; Jinho Kim; Yang-Sae Moon. Springer Verlag, 2017. p. 343-355 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10234 LNAI).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - MC3

T2 - 21st Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2017

AU - Chakraborty, Tanmoy

AU - Chandhok, Des

AU - Subrahmanian, V. S.

N1 - Funding Information: Parts of this work were funded by ONR grants N00014-15-R-BA010, N00014-16-R-BA01, N000141612739 and N000141612918. Publisher Copyright: © 2017, Springer International Publishing AG.

PY - 2017

Y1 - 2017

N2 - In this paper, we propose MC3, an ensemble framework for multi-class classification. MC3 is built on “consensus learning”, a novel learning paradigm where each individual base classifier keeps on improving its classification by exploiting the outcomes obtained from other classifiers until a consensus is reached. Based on this idea, we propose two algorithms, MC3-R and MC3-S that make different trade-offs between quality and runtime. We conduct rigorous experiments comparing MC3-R and MC3-S with 12 baseline classifiers on 13 different datasets. Our algorithms perform as well or better than the best baseline classifier, achieving on average, a 5.56% performance improvement. Moreover, unlike existing baseline algorithms, our algorithms also improve the performance of individual base classifiers up to 10%. (The code is available at https://github.com/MC3-code.).

AB - In this paper, we propose MC3, an ensemble framework for multi-class classification. MC3 is built on “consensus learning”, a novel learning paradigm where each individual base classifier keeps on improving its classification by exploiting the outcomes obtained from other classifiers until a consensus is reached. Based on this idea, we propose two algorithms, MC3-R and MC3-S that make different trade-offs between quality and runtime. We conduct rigorous experiments comparing MC3-R and MC3-S with 12 baseline classifiers on 13 different datasets. Our algorithms perform as well or better than the best baseline classifier, achieving on average, a 5.56% performance improvement. Moreover, unlike existing baseline algorithms, our algorithms also improve the performance of individual base classifiers up to 10%. (The code is available at https://github.com/MC3-code.).

KW - Consensus

KW - Ensemble learning

KW - Multi-class classification

UR - http://www.scopus.com/inward/record.url?scp=85018368052&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85018368052&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-57454-7_27

DO - 10.1007/978-3-319-57454-7_27

M3 - Conference contribution

AN - SCOPUS:85018368052

SN - 9783319574530

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 343

EP - 355

BT - Advances in Knowledge Discovery and Data Mining - 21st Pacific-Asia Conference, PAKDD 2017, Proceedings

A2 - Shim, Kyuseok

A2 - Lee, Jae-Gil

A2 - Cao, Longbing

A2 - Lin, Xuemin

A2 - Kim, Jinho

A2 - Moon, Yang-Sae

PB - Springer Verlag

Y2 - 23 May 2017 through 26 May 2017

ER -

Chakraborty T, Chandhok D, Subrahmanian VS. MC3: A multi-class consensus classification framework. In Shim K, Lee JG, Cao L, Lin X, Kim J, Moon YS, editors, Advances in Knowledge Discovery and Data Mining - 21st Pacific-Asia Conference, PAKDD 2017, Proceedings. Springer Verlag. 2017. p. 343-355. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-57454-7_27

MC3: A multi-class consensus classification framework

Abstract

Publication series

Conference

Funding

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this