DocumentCode :
1521228
Title :
Addressing Missing Values in Kernel-Based Multimodal Biometric Fusion Using Neutral Point Substitution
Author :
Poh, Norman ; Windridge, David ; Mottl, Vadim ; Tatarchuk, Alexander ; Eliseyev, Andrey
Author_Institution :
CVSSP, Univ. of Surrey, Guildford, UK
Volume :
5
Issue :
3
fYear :
2010
Firstpage :
461
Lastpage :
469
Abstract :
In multimodal biometric information fusion, it is common to encounter missing modalities in which matching cannot be performed. As a result, at the match score level, this implies that scores will be missing. We address the multimodal fusion problem involving missing modalities (scores) using support vector machines (SVMs) with the neutral point substitution (NPS) method. The approach starts by processing each modality using a kernel. When a modality is missing, at the kernel level, the missing modality is substituted by one that is unbiased with regards to the classification, called a neutral point. Critically, unlike conventional missing-data substitution methods, explicit calculation of neutral points may be omitted by virtue of their implicit incorporation within the SVM training framework. Experiments based on the publicly available Biosecure DS2 multimodal (scores) data set show that the SVM-NPS approach achieves very good generalization performance compared to the sum rule fusion, especially with severe missing modalities.
Keywords :
biometrics (access control); pattern classification; sensor fusion; support vector machines; SVM training framework; kernel-based multimodal biometric fusion; missing modality; neutral point classification; neutral point substitution method; support vector machines; Authentication; Biometrics; Business; Government; Kernel; Security; Support vector machine classification; Support vector machines; Testing; Throughput; Biometric authentication; information fusion; missing features; multimodal biometrics; multiple classifiers system;
fLanguage :
English
Journal_Title :
Information Forensics and Security, IEEE Transactions on
Publisher :
ieee
ISSN :
1556-6013
Type :
jour
DOI :
10.1109/TIFS.2010.2053535
Filename :
5491168
Link To Document :
بازگشت