Total margin based adaptive fuzzy support vector machines for multiview face recognition

Multiview face recognition is a very difficult pattern recognition problem due to its large variation. And support vector machine (SVM) can serve as a robust classifier for its excellent generalization ability. This paper proposes a new class called total margin based adaptive fuzzy support vector machines (TAF-SVM) to deal with the some problems that may occur in SVM when applied to multiview face recognition. The proposed TAF-SVM not only solves the overfitting problem due to outliers but also corrects the skew of the optimal separating hyperplane due to the training from very imbalanced datasets. In addition, by introducing the total margin algorithm, a lower generalization error bound can be obtained The above three goals are embodied into the traditional SVM so that the TAF-SVM is proposed and reformulated in both linear and nonlinear cases in this paper. By using the CYCU multiview face database and the kernel Fisher's discriminant analysis (KFDA) method to extract discriminating face features, experimental results indicate that the proposed TAF-SVM is superior to the traditional SVM for multiview face recognition. Also, results demonstrate that the proposed TAF-SVM can achieve smaller error variances than SVM.