摘要
提出了基于自适应投影算法和修正核函数算法的混合支撑矢量机 ,根据修正核函数算法的正形投影变换 ,将问题映射到黎曼空间来增大分类面 ,从而提高支撑矢量机的分类精度 .但其缺陷在于它是由两步优化实现的 ,因此增加了时间复杂度 .基于此 ,混合支撑矢量机使用自适应投影算法对支撑矢量进行预选取 ,即通过投影从训练样本中选择部分样本作为中心矢量进行训练 .实验结果表明
A novel hybrid support vector machine based on the adaptive projective algorithm and modifying kernel functions method is proposed. The modifying kernel functions method can improve the performance of a support vector machine classifier by a conformal mapping, but the remarkable decrease in speed is its serious disadvantage. Therefore, the hybrid support vector machine is used to pre-extract support vectors by the adaptive projective algorithm, so as to train the center vector chosen from the training data. It is shown that the separability between classes is raised and speed is greatly increased with this hybrid support vector machine.
出处
《西安电子科技大学学报》
EI
CAS
CSCD
北大核心
2002年第4期477-481,共5页
Journal of Xidian University
基金
国家自然科学基金资助项目 (60 0 73 0 5 3 )
