Apply an Adaptive Center Selection Algorithm to Radial Basis Function Neural Network for Face Recognition

In general, the principal component analysis (PCA) technique is applied to reduce the feature dimensions. In this paper, different from traditional PCAs, the PCA is used to select adequate centers for the classifier of radial basis function neural networks (RBFNN). In addition, a novel weights updating method is also included in the RBFNN for face recognition. The specific design, not only increases the convergent speed, but also retains generalization ability. Experimental results show the proposed method has high recognition rate with a short training time.