Dual uncertainty minimization regularization and its applications on heterogeneous data

Yu Cheng, Alok Choudhary, Jun Wang, Sharath Pankanti, Huan Liu

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In many practical machine learning systems, the prediction/classification tasks involve the usage of heterogeneous data in semi-supervised settings, where the objective is to maximize the utility of multiple views (usually dual views) information from the data. In this work, we propose a general framework, Dual Uncertainty Minimization Regularization (DUMR), that maximizes the usage of heterogeneous data for a dual view semi-supervised classification/prediction. Through extending a recent uncertainty regularizer to a heterogeneous setting, we propose to optimize an objective which ensures the minimum uncertainty of the prediction over both views extracted from heterogeneous source. In specific, for different problem settings, we design two type of uncertainty regularizer with entropy and squared-loss mutual information, separately. The proposed framework is exploited in three datamining/multimeida analysis tasks, social role identification, legislative prediction and action recognition, and the comparison with other peer methods corroborate the superior performance of the proposed method.

Original languageEnglish (US)
Title of host publicationIEEE International Conference on Data Mining Workshops, ICDMW
PublisherIEEE Computer Society
Pages1163-1170
Number of pages8
Volume2015-January
EditionJanuary
DOIs
StatePublished - Jan 26 2015
Event14th IEEE International Conference on Data Mining Workshops, ICDMW 2014 - Shenzhen, China
Duration: Dec 14 2014 → …

Other

Other14th IEEE International Conference on Data Mining Workshops, ICDMW 2014
CountryChina
CityShenzhen
Period12/14/14 → …

Fingerprint

Learning systems
Entropy
Uncertainty

Keywords

  • Dual Uncertainty Minimization
  • Heterogeneous Data
  • Multiple-Views Learning

ASJC Scopus subject areas

  • Computer Science Applications
  • Software

Cite this

Cheng, Y., Choudhary, A., Wang, J., Pankanti, S., & Liu, H. (2015). Dual uncertainty minimization regularization and its applications on heterogeneous data. In IEEE International Conference on Data Mining Workshops, ICDMW (January ed., Vol. 2015-January, pp. 1163-1170). [7022727] IEEE Computer Society. https://doi.org/10.1109/ICDMW.2014.138

Dual uncertainty minimization regularization and its applications on heterogeneous data. / Cheng, Yu; Choudhary, Alok; Wang, Jun; Pankanti, Sharath; Liu, Huan.

IEEE International Conference on Data Mining Workshops, ICDMW. Vol. 2015-January January. ed. IEEE Computer Society, 2015. p. 1163-1170 7022727.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Cheng, Y, Choudhary, A, Wang, J, Pankanti, S & Liu, H 2015, Dual uncertainty minimization regularization and its applications on heterogeneous data. in IEEE International Conference on Data Mining Workshops, ICDMW. January edn, vol. 2015-January, 7022727, IEEE Computer Society, pp. 1163-1170, 14th IEEE International Conference on Data Mining Workshops, ICDMW 2014, Shenzhen, China, 12/14/14. https://doi.org/10.1109/ICDMW.2014.138
Cheng Y, Choudhary A, Wang J, Pankanti S, Liu H. Dual uncertainty minimization regularization and its applications on heterogeneous data. In IEEE International Conference on Data Mining Workshops, ICDMW. January ed. Vol. 2015-January. IEEE Computer Society. 2015. p. 1163-1170. 7022727 https://doi.org/10.1109/ICDMW.2014.138
Cheng, Yu ; Choudhary, Alok ; Wang, Jun ; Pankanti, Sharath ; Liu, Huan. / Dual uncertainty minimization regularization and its applications on heterogeneous data. IEEE International Conference on Data Mining Workshops, ICDMW. Vol. 2015-January January. ed. IEEE Computer Society, 2015. pp. 1163-1170
@inproceedings{90a05b804e46410284e4e33203f46a37,
title = "Dual uncertainty minimization regularization and its applications on heterogeneous data",
abstract = "In many practical machine learning systems, the prediction/classification tasks involve the usage of heterogeneous data in semi-supervised settings, where the objective is to maximize the utility of multiple views (usually dual views) information from the data. In this work, we propose a general framework, Dual Uncertainty Minimization Regularization (DUMR), that maximizes the usage of heterogeneous data for a dual view semi-supervised classification/prediction. Through extending a recent uncertainty regularizer to a heterogeneous setting, we propose to optimize an objective which ensures the minimum uncertainty of the prediction over both views extracted from heterogeneous source. In specific, for different problem settings, we design two type of uncertainty regularizer with entropy and squared-loss mutual information, separately. The proposed framework is exploited in three datamining/multimeida analysis tasks, social role identification, legislative prediction and action recognition, and the comparison with other peer methods corroborate the superior performance of the proposed method.",
keywords = "Dual Uncertainty Minimization, Heterogeneous Data, Multiple-Views Learning",
author = "Yu Cheng and Alok Choudhary and Jun Wang and Sharath Pankanti and Huan Liu",
year = "2015",
month = "1",
day = "26",
doi = "10.1109/ICDMW.2014.138",
language = "English (US)",
volume = "2015-January",
pages = "1163--1170",
booktitle = "IEEE International Conference on Data Mining Workshops, ICDMW",
publisher = "IEEE Computer Society",
edition = "January",

}

TY - GEN

T1 - Dual uncertainty minimization regularization and its applications on heterogeneous data

AU - Cheng, Yu

AU - Choudhary, Alok

AU - Wang, Jun

AU - Pankanti, Sharath

AU - Liu, Huan

PY - 2015/1/26

Y1 - 2015/1/26

N2 - In many practical machine learning systems, the prediction/classification tasks involve the usage of heterogeneous data in semi-supervised settings, where the objective is to maximize the utility of multiple views (usually dual views) information from the data. In this work, we propose a general framework, Dual Uncertainty Minimization Regularization (DUMR), that maximizes the usage of heterogeneous data for a dual view semi-supervised classification/prediction. Through extending a recent uncertainty regularizer to a heterogeneous setting, we propose to optimize an objective which ensures the minimum uncertainty of the prediction over both views extracted from heterogeneous source. In specific, for different problem settings, we design two type of uncertainty regularizer with entropy and squared-loss mutual information, separately. The proposed framework is exploited in three datamining/multimeida analysis tasks, social role identification, legislative prediction and action recognition, and the comparison with other peer methods corroborate the superior performance of the proposed method.

AB - In many practical machine learning systems, the prediction/classification tasks involve the usage of heterogeneous data in semi-supervised settings, where the objective is to maximize the utility of multiple views (usually dual views) information from the data. In this work, we propose a general framework, Dual Uncertainty Minimization Regularization (DUMR), that maximizes the usage of heterogeneous data for a dual view semi-supervised classification/prediction. Through extending a recent uncertainty regularizer to a heterogeneous setting, we propose to optimize an objective which ensures the minimum uncertainty of the prediction over both views extracted from heterogeneous source. In specific, for different problem settings, we design two type of uncertainty regularizer with entropy and squared-loss mutual information, separately. The proposed framework is exploited in three datamining/multimeida analysis tasks, social role identification, legislative prediction and action recognition, and the comparison with other peer methods corroborate the superior performance of the proposed method.

KW - Dual Uncertainty Minimization

KW - Heterogeneous Data

KW - Multiple-Views Learning

UR - http://www.scopus.com/inward/record.url?scp=84936878792&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84936878792&partnerID=8YFLogxK

U2 - 10.1109/ICDMW.2014.138

DO - 10.1109/ICDMW.2014.138

M3 - Conference contribution

VL - 2015-January

SP - 1163

EP - 1170

BT - IEEE International Conference on Data Mining Workshops, ICDMW

PB - IEEE Computer Society

ER -