Addressing Missing Values in Kernel-Based Multimodal Biometric Fusion Using Neutral Point Substitution

摘要: 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 missing. We address fusion problem involving (scores) using support vector machines (SVMs) with neutral point substitution (NPS) method. The approach starts by processing each modality kernel. When missing, kernel substituted one unbiased regards classification, called point. Critically, unlike conventional missing-data methods, explicit calculation of points may omitted virtue their implicit incorporation within SVM training framework. Experiments based on publicly available Biosecure DS2 data set show SVM-NPS achieves very good generalization performance compared sum rule especially severe modalities.