【摘要】 提出基于模式训练的Voronoi划分向量分类器结构,指出该结构是以欧几里得距离为测量的最优模式分类器,给出M类k维(k≥2)Voronoi域分类器的两个判决准则和三种结构。更多还原

【Abstract】 A new structure of vector classifiers, based upon space partition of voronot region according to a pattern set gem-rated by classed patterns training, is presented. The fact that this structure is an optimal pattern classifier on Euclidean distance is shown. Two decision rules and three kinds of M classes and k-dimension classifier structures on voronoi region are given.更多还原