PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Non-Linear Variance Reduction Techniques in Biometric Authentication
Norman Poh and Samy Bengio
In: Mulitmodal User Authentication Workshop (MMUA) 2003, 11-12 December 2003, Santa Barbara, California.


In this paper, several approaches that can be used to improve biometric authentication applications are proposed. The idea is inspired by the ensemble approach, i.e., the use of several classifiers to solve a problem. Compared to using only one classifier, the ensemble of classifiers has the advantage of reducing the overall variance of the system. Instead of using multiple classifiers, we propose here to examine other possible means of variance reduction (VR), namely through the use of multiple synthetic samples, different extractors (features) and biometric modalities. The scores are combined using the average operator, Multi-Layer Perceptron and Support Vector Machines. It is found empirically that VR via modalities is the best technique, followed by VR via extractors, VR via classifiers and VR via synthetic samples. This order of effectiveness is due to the corresponding degree of independence of the combined objects (in decreasing order). The theoretical and empirical findings show that the combined experts via VR techniques \emph{always} perform better than the average of their participating experts. Furthermore, in practice, \emph{most} combined experts perform better than any of their participating experts.

PDF - Requires Adobe Acrobat Reader or other PDF viewer.
EPrint Type:Conference or Workshop Item (Paper)
Additional Information:biometric authentication fusion
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Multimodal Integration
ID Code:820
Deposited By:Norman Poh
Deposited On:01 January 2005