Show simple item record

dc.contributor.authorPham, DucSon
dc.contributor.authorVenkatesh, Svetha
dc.contributor.editorD Wang
dc.contributor.editorM Reynolds
dc.date.accessioned2017-01-30T15:00:04Z
dc.date.available2017-01-30T15:00:04Z
dc.date.created2012-03-01T20:00:56Z
dc.date.issued2011
dc.identifier.citationPham, Duc-Son and Venkatesh, Svetha. 2011. Supervised subspace learning with multi-class Lagrangian SVM on the Grassmann Manifold, in Wang, Dianhui and Reynolds, Mark (ed), AI 2011: Advances in Artificial Intelligence: 24th Australasian Joint Conference, Dec 5-8 2011, pp. 241-250. Perth, Australia: Springer
dc.identifier.urihttp://hdl.handle.net/20.500.11937/42511
dc.identifier.doi10.1007/978-3-642-25832-9_25
dc.description.abstract

Learning robust subspaces to maximize class discrimination is challenging, and most current works consider a weak connection between dimensionality reduction and classifier design. We propose an alternate framework wherein these two steps are combined in a joint formulation to exploit the direct connection between dimensionality reduction and classification. Specifically, we learn an optimal subspace on the Grassmann manifold jointly minimizing the classification error of an SVM classifier. We minimize the regularized empirical risk over both the hypothesis space of functions that underlies this new generalized multiclass Lagrangian SVM and the Grassmann manifold such that a linear projection is to be found. We propose an iterative algorithm to meet the dual goal of optimizing both the classifier and projection. Extensive numerical studies on challenging datasets show robust performance of the proposed scheme over other alternatives in contexts wherein limited training data is used, verifying the advantage of the joint formulation.

dc.publisherSpringer
dc.titleSupervised subspace learning with multi-class Lagrangian SVM on the Grassmann Manifold
dc.typeConference Paper
dcterms.source.startPage241
dcterms.source.endPage250
dcterms.source.titleAI 2011: Advances in Artificial Intelligence
dcterms.source.seriesAI 2011: Advances in Artificial Intelligence
dcterms.source.isbn9783642258312
dcterms.source.conferenceAI 2011
dcterms.source.conference-start-dateDec 5 2011
dcterms.source.conferencelocationPerth, Australia
dcterms.source.placeUSA
curtin.departmentDepartment of Computing
curtin.accessStatusFulltext not available


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record