多类扩码支撑矢量机  

Multi-class SVM Based On Extended Code

作　　者：任双桥[1] 王宏强[1] 黎湘[1] 庄钊文[1] 

机构地区：[1]国防科技大学电子科学与工程学院空间电子信息技术研究所,湖南长沙410073

出　　处：《信号处理》2009年第6期916-924,共9页Journal of Signal Processing

基　　金：国家自然科学基金(No.60402032);国防预研项目(No.41303040203)

摘　　要：本文通过扩展类别标示符二进制编码,提升决策空间的维数以增强决策函数的分类能力,提出了一种新的多类扩码支撑矢量机模型——半对半算法(Half-Versus-Half)。该模型通过序贯解耦求解,有效地提高了运算速度,同时克服了测试过程中的判决"死区"问题。另外,基于统计学习理论,本文还分析和探讨了多类目标扩码识别算法的推广性能。实测数据实验结果表明,半对半算法在计算速度和识别性能上有了明显的提高。This paper puts forward a multi-class SVM based on extending the dimension of decision space, which is called Half- Versus-Half （H-V-H） algorithm. The H-V-H model can be sequentially solved and has no decision blind region. In addition, the paper has analyzed the generalized performance of multi-class objection recognition algorithm based on extended code. As the experimental re- sults shown, validated with the actually measuring data, the performance of the H-V-H method, such as classification capability and generalized capability are improved obviously in contrast to the classical method.

关 键 词：多类目标 支撑矢量机 扩码 半对半算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]