Location and scatter matching for dataset shift in text mining

Chen, B; Lam, W; Tsang, I; Wong, TL

Location and scatter matching for dataset shift in text mining

Chen, B Lam, W Tsang, I

Wong, TL

Permalink

Publication Type:: Conference Proceeding
Citation:: Proceedings - IEEE International Conference on Data Mining, ICDM, 2010, pp. 773 - 778
Issue Date:: 2010-12-01

Closed Access

	Filename	Description	Size
	2013006912OK.pdf	Published version	220.22 kB	Adobe PDF	View/Open

Copyright Clearance Process

Recently Added
In Progress
Closed Access

This item is closed access and not available.

Full metadata record

Field	Value	Language
dc.contributor.author	Chen, B	en_US
dc.contributor.author	Lam, W	en_US
dc.contributor.author	Tsang, I https://orcid.org/0000-0001-8095-4637	en_US
dc.contributor.author	Wong, TL	en_US
dc.date.issued	2010-12-01	en_US
dc.identifier.citation	Proceedings - IEEE International Conference on Data Mining, ICDM, 2010, pp. 773 - 778	en_US
dc.identifier.isbn	9780769542560	en_US
dc.identifier.issn	1550-4786	en_US
dc.identifier.uri	http://hdl.handle.net/10453/32148
dc.description.abstract	Dataset shift from the training data in a source domain to the data in a target domain poses a great challenge for many statistical learning methods. Most algorithms can be viewed as exploiting only the first-order statistics, namely, the empirical mean discrepancy to evaluate the distribution gap. Intuitively, considering only the empirical mean may not be statistically efficient. In this paper, we propose a non-parametric distance metric with a good property which jointly considers the empirical mean (Location) and sample covariance (Scatter) difference. More specifically, we propose an improved symmetric Stein's loss function which combines the mean and covariance discrepancy into a unified Bregman matrix divergence of which Jensen-Shannon divergence between normal distributions is a particular case. Our target is to find a good feature representation which can reduce the distribution gap between different domains, at the same time, ensure that the new derived representation can encode most discriminative components with respect to the label information. We have conducted extensive experiments on several document classification datasets to demonstrate the effectiveness of our proposed method. © 2010 IEEE.	en_US
dc.relation.ispartof	Proceedings - IEEE International Conference on Data Mining, ICDM	en_US
dc.relation.isbasedon	10.1109/ICDM.2010.72	en_US
dc.title	Location and scatter matching for dataset shift in text mining	en_US
dc.type	Conference Proceeding
utslib.for	0801 Artificial Intelligence and Image Processing	en_US
dc.location.activity	Sydney, Australia
pubs.embargo.period	Not known	en_US
pubs.organisational-group	/University of Technology Sydney
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology
pubs.organisational-group	/University of Technology Sydney/Strength - CAI - Centre for Artificial Intelligence
utslib.copyright.status	closed_access
pubs.publication-status	Published	en_US

Abstract:

Dataset shift from the training data in a source domain to the data in a target domain poses a great challenge for many statistical learning methods. Most algorithms can be viewed as exploiting only the first-order statistics, namely, the empirical mean discrepancy to evaluate the distribution gap. Intuitively, considering only the empirical mean may not be statistically efficient. In this paper, we propose a non-parametric distance metric with a good property which jointly considers the empirical mean (Location) and sample covariance (Scatter) difference. More specifically, we propose an improved symmetric Stein's loss function which combines the mean and covariance discrepancy into a unified Bregman matrix divergence of which Jensen-Shannon divergence between normal distributions is a particular case. Our target is to find a good feature representation which can reduce the distribution gap between different domains, at the same time, ensure that the new derived representation can encode most discriminative components with respect to the label information. We have conducted extensive experiments on several document classification datasets to demonstrate the effectiveness of our proposed method. © 2010 IEEE.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/32148