摘要
支持向量机采用二次规划提取支持向量,计算量大,所需内存大,速度慢,在识别应用上影响识别速度。因此从另外的角度出发,利用空间转化,将原样本空间转化为另一便于识别的空间,在另一空间中运用线性规划进行样本的拟支持向量的提取,再用拟支持向量对应的判决向量进行识别。实验表明算法适合大样本识别,可进行飞机目标识别,而且精度很高。与支持向量机相比,拟支持向量需要的内存少,识别时间相对短,识别精度高等优点。
The support vector machine uses two - time programming to withdraw the support vector, which has big computation quantity, needs huge memory, has low speed, thus affecting.the recognition speed, This paper changes the space of original samples into another space in which it is easy to recognize targets. In this space, linear programming is used to extract simulative support vector of the samples, then the decision vector corresponding to simu- lative support vector is used to recognize targets. Results of experiments show that this arithmetic is suited to recog-nize a large amount of samples, It can recognize aircraft targets and the recognition precision is very high. Comparing with support vector machine, simulative support vector needs less EMS memory, its recognition time is shorter and recognition precision is very high.
出处
《计算机仿真》
CSCD
2008年第2期100-103,共4页
Computer Simulation
关键词
线性规划
拟支持向量
目标识别
Linear programming
Simulative support vector
Target recognition