Biometrical Letters Vol. 51(1), 2014, pp. 57-73
|
A KERNEL-BASED LEARNING ALGORITHM COMBINING KERNEL DISCRIMINANT COORDINATES AND KERNEL PRINCIPAL COMPONENTS Karol Deręgowski1, Mirosław Krzyśko2 1President Stanislaw Wojciechowski Higher Vocational State School in Kalisz, Institute of Management, Nowy Świat 4, 62-800 Kalisz, Poland, e-mail: k.deregowski@pwsz.kalisz.pl 2Adam Mickiewicz University, Faculty of Mathematics and Computer Science, Umultowska 87, 61-614 Poznań, Poland, e-mail: mkrzysko@amu.edu.pl |
Kernel principal components (KPC) and kernel discriminant coordinates (KDC), which are the extensions of principal components and discriminant coordinates, respectively, from a linear domain to a nonlinear domain via the kernel trick, are two very popular nonlinear feature extraction methods. The kernel discriminant coordinates space has proven to be a very powerful space for pattern recognition. However, further study shows that there are still drawbacks in this method. To improve the performance of pattern recognition, we propose a new learning algorithm combining the advantages of KPC and KDC.
kernel principal components, kernel discriminant coordinates
