A Hybrid Model Combining Learning Distance Metric and DAG Support Vector Machine for Multimodal Biometric Recognition

Metric learning has significantly improved machine applications such as face re-identification and image classification using K-Nearest Neighbor (KNN) Support Vector Machine (SVM) classifiers. However, to the best of our knowledge, it not been investigated yet, especially for multimodal biometric recognition problem in immigration, forensic surveillance with uncontrolled ear datasets. Therefore, is interesting very attractive propose a novel framework based on Learning Distance (LDM) via kernel SVM. This paper considers metric SVM by investigating hybrid Directed Acyclic Graph (LDM-DAGSVM) model recognition, where LDM DAGSVM are two emerging techniques dealing problems. Different from existing methods, proposed approach aims learn Mahalanobis distance maximize inter-class variations minimize intra-class variations, simultaneously. Experimental results datasets AR AWE show that achieves competitive performance compared models working individual modalities overperforms state-of-the-art methods. The five-fold accuracy around 99.85 % images.