Copula Based Classifier Fusion under Statistical Dependence

Onur Ozdemir; Thomas G. Allen; Sora Choi; Thakshila Wimalajeewa; Pramod K. Varshney

doi:10.1109/TPAMI.2017.2774300

Copula Based Classifier Fusion under Statistical Dependence

Onur Ozdemir, Thomas G. Allen, Sora Choi, Thakshila Wimalajeewa, Pramod K. Varshney

Department of Electrical Engineering & Computer Science

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

16 Scopus citations

Abstract

We consider the problem of fusing probability scores from a set of classifiers to estimate a final fused probability score. Our interest is in scenarios where the classifiers are statistically dependent. To that end, we propose a new classifier fusion approach that is data driven and founded on the statistical theory of copulas. Numerical results with both simulated and real data show that our copula based classifier fusion approach produces better probability scores than individual classifiers and outperforms existing probability score fusion approaches.

Original language	English (US)
Article number	8113592
Pages (from-to)	2740-2748
Number of pages	9
Journal	IEEE Transactions on Pattern Analysis and Machine Intelligence
Volume	40
Issue number	11
DOIs	https://doi.org/10.1109/TPAMI.2017.2774300
State	Published - Nov 1 2018

Keywords

Copulas
classification
probability score fusion
statistical dependence

ASJC Scopus subject areas

Software
Computer Vision and Pattern Recognition
Computational Theory and Mathematics
Artificial Intelligence
Applied Mathematics

Access to Document

10.1109/TPAMI.2017.2774300

Cite this

@article{539c2beb992b4a33bc4ce762723a05d4,

title = "Copula Based Classifier Fusion under Statistical Dependence",

abstract = "We consider the problem of fusing probability scores from a set of classifiers to estimate a final fused probability score. Our interest is in scenarios where the classifiers are statistically dependent. To that end, we propose a new classifier fusion approach that is data driven and founded on the statistical theory of copulas. Numerical results with both simulated and real data show that our copula based classifier fusion approach produces better probability scores than individual classifiers and outperforms existing probability score fusion approaches.",

keywords = "Copulas, classification, probability score fusion, statistical dependence",

author = "Onur Ozdemir and Allen, {Thomas G.} and Sora Choi and Thakshila Wimalajeewa and Varshney, {Pramod K.}",

note = "Funding Information: The authors would like to thank Elizabeth Benagh and Francis O{\textquoteright}Donovan for their assistance with the collection of Twitter data and extraction of user features. This work was in part sponsored by the DARPA SMISC program. Approved for Public Release, 16-MDA-8938 (8 December 16). Publisher Copyright: {\textcopyright} 1979-2012 IEEE.",

year = "2018",

month = nov,

day = "1",

doi = "10.1109/TPAMI.2017.2774300",

language = "English (US)",

volume = "40",

pages = "2740--2748",

journal = "IEEE Transactions on Pattern Analysis and Machine Intelligence",

issn = "0162-8828",

publisher = "IEEE Computer Society",

number = "11",

}

TY - JOUR

T1 - Copula Based Classifier Fusion under Statistical Dependence

AU - Ozdemir, Onur

AU - Allen, Thomas G.

AU - Choi, Sora

AU - Wimalajeewa, Thakshila

AU - Varshney, Pramod K.

N1 - Funding Information: The authors would like to thank Elizabeth Benagh and Francis O’Donovan for their assistance with the collection of Twitter data and extraction of user features. This work was in part sponsored by the DARPA SMISC program. Approved for Public Release, 16-MDA-8938 (8 December 16). Publisher Copyright: © 1979-2012 IEEE.

PY - 2018/11/1

Y1 - 2018/11/1

N2 - We consider the problem of fusing probability scores from a set of classifiers to estimate a final fused probability score. Our interest is in scenarios where the classifiers are statistically dependent. To that end, we propose a new classifier fusion approach that is data driven and founded on the statistical theory of copulas. Numerical results with both simulated and real data show that our copula based classifier fusion approach produces better probability scores than individual classifiers and outperforms existing probability score fusion approaches.

AB - We consider the problem of fusing probability scores from a set of classifiers to estimate a final fused probability score. Our interest is in scenarios where the classifiers are statistically dependent. To that end, we propose a new classifier fusion approach that is data driven and founded on the statistical theory of copulas. Numerical results with both simulated and real data show that our copula based classifier fusion approach produces better probability scores than individual classifiers and outperforms existing probability score fusion approaches.

KW - Copulas

KW - classification

KW - probability score fusion

KW - statistical dependence

UR - http://www.scopus.com/inward/record.url?scp=85035129035&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85035129035&partnerID=8YFLogxK

U2 - 10.1109/TPAMI.2017.2774300

DO - 10.1109/TPAMI.2017.2774300

M3 - Article

C2 - 29990102

AN - SCOPUS:85035129035

VL - 40

SP - 2740

EP - 2748

JO - IEEE Transactions on Pattern Analysis and Machine Intelligence

JF - IEEE Transactions on Pattern Analysis and Machine Intelligence

SN - 0162-8828

IS - 11

M1 - 8113592

ER -

Copula Based Classifier Fusion under Statistical Dependence

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this