Abstract:

Algorithm total margin v minimum class variance support vector machines based on common vectors (TM-v-M(CV)²SVMs) were presented for noisy face recognition, which integrates the advantages of minimum class variance support vector machines(MCVSVMs)and total margin v support vector machine(TM-v-SVM). Based on common vectors (CVs), the divergence matrix was introduced to improve the classification and anti-noisy performances of noisy face classification, and TM-v-M (CV)²SVMs derivation was given. The experimental results about noisy face classification showed that the proposed TM-v-M(CV)²SVMs had better classification performance than both the MCVSVMs and TM-v-SVM.

Key words:  support vector machine; minimum class variance support vector machines; total margin v support vector machine; discriminative common vectors; common vectors; face recognition