An efficient and effective method to solve kernel - Fisher discriminant analysis

In this paper, an efficient and effective method to solve kernel Fisher discriminant analysis is proposed. Since the QR decomposition on the small-size matrix is adopted, the superiority of the proposed method is its computational efficiency. Moreover, the proposed method can avoid the singularity problem. Most importantly, the proposed method shows that the maximal number of kernel discriminant vectors is the same as that of linear discriminant vectors. Experimental results on handwritten numeral characters show that the proposed method is effective and feasible. (C) 2004 Elsevier B.V. All rights reserved.