摘要
普通的最小二乘支持向量机(LS-SVM)稀疏化算法在处理有些常见的模式识别问题时,随着训练样本的删减,识别率下滑很快,往往达不到稀疏化的目的。针对这种情况,提出了一种新的LS-SVM稀疏化算法来弥补这种不足,从而使得LS-SVM稀疏化算法体系更加完善。将新算法应用到雷达一维距离像的识别中,实验结果证明了新算法的有效性。
The recognition rate of Least Squares Support Vector Machine （LS-SVM） sparse algorithm rapidly decreases with the reduction of training samples in dealing with some pattern recognition issues, and the sparsifieation can not be achieved. To overcome such a shortage, a new sparse algorithm was proposed. The method was applied to radar range profile＇s recognition and the experimental results show its validity in recognition.
出处
《计算机应用》
CSCD
北大核心
2009年第6期1559-1562,1581,共5页
journal of Computer Applications
关键词
最小二乘支持向量机
稀疏化
雷达一维距离像
Least Squares Support Vector Machine （LS-SVM）
sparsification
radar range profile
