Reference : From subspace learning to distance learning: a geometrical optimization approach
Scientific congresses and symposiums : Paper published in a book
Engineering, computing & technology : Computer science
http://hdl.handle.net/2268/30476
From subspace learning to distance learning: a geometrical optimization approach
English
Meyer, Gilles mailto [Université de Liège - ULiège > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Systèmes et modélisation >]
Journée, Michel [Université de Liège - ULiège > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Systèmes et modélisation > >]
Bonnabel, Silvère mailto [Mines ParisTech > Robotic Center (CAOR) > Mathématiques et Systèmes > >]
Sepulchre, Rodolphe mailto [Université de Liège - ULiège > Dép. d'électric., électron. et informat. (Inst.Montefiore) > Systèmes et modélisation >]
2009
Proceedings of the 2009 IEEE Workshop on Statistical Signal Processing (SSP2009)
385 - 388
No
Yes
International
978-1-4244-2709-3
IEEE Workshop on Statistical Signal Processing (SSP2009)
du 31 août 2009 au 3 septembre 2009
Cardiff
Wales
[en] manifold-based optimization ; kernel and metric learning ; low-rank approximation ; online learning
[en] In this paper, we adopt a differential-geometry viewpoint to tackle the problem of learning a distance online. As this prob- lem can be cast into the estimation of a fixed-rank positive semidefinite (PSD) matrix, we develop algorithms that ex- ploits the rich geometry structure of the set of fixed-rank PSD matrices. We propose a method which separately updates the subspace of the matrix and its projection onto that subspace. A proper weighting of the two iterations enables to continu- ously interpolate between the problem of learning a subspace and learning a distance when the subspace is fixed.
Fonds de la Recherche Scientifique (Communauté française de Belgique) - F.R.S.-FNRS
Researchers ; Professionals
http://hdl.handle.net/2268/30476
10.1109/SSP.2009.5278557

File(s) associated to this reference

Fulltext file(s):

FileCommentaryVersionSizeAccess
Open access
ssp09.pdfAuthor preprint69.96 kBView/Open

Bookmark and Share SFX Query

All documents in ORBi are protected by a user license.