Title :
Local Fisher embedding
Author :
De Ridder, Dick ; Loog, Marco ; Reinders, Marcel J T
Author_Institution :
Dept. of Electr. Eng. & Mech. Comput. Sci., Delft Univ. of Technol., Netherlands
Abstract :
In recent work, several supervised spectral embedding procedures have been proposed. Although experimentally validated, the construction of these algorithms was rather ad hoc. This paper shows how supervised locally linear embedding can be seen as performing a local Fisher mapping. A new formulation, combining linear and Fisher embedding, is proposed and experimentally validated.
Keywords :
data visualisation; pattern recognition; local Fisher embedding; supervised locally linear embedding; supervised spectral embedding; Assembly; Biomedical imaging; Character generation; Data visualization; Feature extraction; Laplace equations; Least squares approximation; Linear discriminant analysis; Pattern recognition; Vectors;
Conference_Titel :
Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference on
Print_ISBN :
0-7695-2128-2
DOI :
10.1109/ICPR.2004.1334176