Graduate School and Research Center in Digital Sciences

Multi-view semi-supervised discriminant analysis: A new approach to audio-visual person recognition

Zhao, Xuran; Evans, Nicholas; Dugelay, Jean-Luc

EUSIPCO 2012, European Signal Processing Conference, August, 27-31, 2012, Bucharest, Romania

Many state-of-the-art biometric systems use feature vectors of high dimension and call for dimensionality reduction techniques to avoid the co-called 'curse of dimensionality.' Supervised approaches such as Linear Discriminant Analysis can extract discriminative features and is used widely, but suffers from over-fitting when used with small datasets. Through the imposition of local adjacency constraints, semisupervised dimensionality reduction techniques can make use of abundant, unlabelled data to improve classification performance. This paper reports a new multi-view, semisupervised discriminant analysis (MSDA) algorithm and its application in audio-visual person recognition. In contrast to existing approaches which typically utilize a single view, MSDA determines a more reliable neighbourhood constraint built jointly from multiple views of the same data. Experimental results on the standard MOBIO database show that our algorithm not only outperforms baseline supervised and unsupervised methods, but that it also outperforms single-view semi-supervised dimension reduction techniques in single view.

Document Bibtex

Title:Multi-view semi-supervised discriminant analysis: A new approach to audio-visual person recognition
Keywords:Audio-visual person recognition, semisupervised learning, discriminant analysis
Type:Conference
Language:English
City:Bucharest
Country:ROMANIA
Date:
Department:Digital Security
Eurecom ref:3727
Copyright: © EURASIP. Personal use of this material is permitted. The definitive version of this paper was published in EUSIPCO 2012, European Signal Processing Conference, August, 27-31, 2012, Bucharest, Romania and is available at :
Bibtex: @inproceedings{EURECOM+3727, year = {2012}, title = {{M}ulti-view semi-supervised discriminant analysis: {A} new approach to audio-visual person recognition}, author = {{Z}hao, {X}uran and {E}vans, {N}icholas and {D}ugelay, {J}ean-{L}uc}, booktitle = {{EUSIPCO} 2012, {E}uropean {S}ignal {P}rocessing {C}onference, {A}ugust, 27-31, 2012, {B}ucharest, {R}omania}, address = {{B}ucharest, {ROMANIA}}, month = {08}, url = {http://www.eurecom.fr/publication/3727} }
See also: