[1]吴 青,刘三阳,郑 巍.基于乘性规则的支持向量机[J].智能系统学报,2007,2(2):74-77.
WU Qing,LIU San-yang,ZHENG Wei.Support vector machines based on multiplicative updates[J].CAAI Transactions on Intelligent Systems,2007,2(2):74-77.
点击复制
《智能系统学报》[ISSN1673-4785/CN23-1538/TP] 卷:
2
期数:
2007年2期
页码:
74-77
栏目:
学术论文—智能系统
出版日期:
2007-04-25
- Title:
-
Support vector machines based on multiplicative updates
- 文章编号:
-
1673-4785(2007)02-0074-04
- 作者:
-
吴 青,刘三阳,郑 巍
-
西安电子科技大学理学院 陕西 西安 710071
- Author(s):
-
WU Qing, LIU San-yang, ZHENG Wei
-
School of Science, Xidian University, Xi’a n 710071, China
-
- 关键词:
-
支持向量机; 二次凸规划; 混合约束; 乘性规则
- Keywords:
-
support vector machine; quadratic convex programming; mixed constraint ; multiplicative update
- 分类号:
-
TP18
- 文献标志码:
-
A
- 摘要:
-
传统的二次规划由于涉及大量的矩阵运算,运算速度慢成为支持向量机的最大缺点. 已有的乘性规则仅适于非负二次凸规划问题,推导出了求解支持向量机中混合约束二次凸规划的乘性规则,利用这一乘性规则极大地提高了优化速度.该方法提供了一种直接优化的方法,其所有变量可以并行迭代,乘性规则可以使得二次规划的目标函数单调下降到它的全局最小点.仿真试验结果表明了该算法有效性.
- Abstract:
-
Due to intensive matrix computation, the speed of the quadratic equati on remains slow. The multiplicative updates available are only suited for nonneg ative quadratic convex programming. In this article, the multiplicative updates are derived for mixed constraint optimizations, dramatically speeding up optimiz ation rate. This method provides an extremely straightforward way to implement s upport vector machines(SVMs) where all the variables can be iterated in parallel. The multiplicative updates converge to global minimum point by monoton ically reducing the target function of quadratic programming. Experimental resul ts have confirmed the effectiveness of our approach.
更新日期/Last Update:
2009-05-06