A KFDA Based on Regularization Method for Multi-classification

Kernel Fisher discriminant analysis (KFDA) improves greatly the multi-classification accuracy of FDA via using kernel trick. The optimal kernel Fisher projection of KFDA can be expressed as a generalized characteristic equation. However, solving the characteristic equation is very difficult, then regularization method is used for it. In this paper, we develop a novel approach to perform regularization parameter based on numerical analysis method. The approach exploits the optimal regularization selection of KFDA to obtain the better classification results. The method is also simple and not computationally complex. Experimental results illustrate the effectiveness of the method.