A Classification Method Using Weighted Nearest Neighbor Method based on A Supervised Decision Algorithm of Plural Class Presentation Points
Hiroki Imamura, Makoto Fujimura, Hideo Kuroda · 2006
To precisely classify input patterns, we propose a pattern classification method which uses weighted distance, based on the mean of norm, from a input pattern to plural class presentation points. By using the proposed method, we can reduce calculation cost of classification and can obtain precise results of classification