一种基于半监督学习的窄带雷达目标识别系统  

A Narrow-Band Radar Target Recognition System Based on Semi-Supervised Learning

作　　者：吴剑旗[1] 田西兰[1] 

机构地区：[1]中国电子科技集团公司第38研究所,合肥230088

出　　处：《中国电子科学研究院学报》2015年第1期49-53,101,共6页Journal of China Academy of Electronics and Information Technology

摘　　要：半监督学习算法利用少量的标注样本与大量的未标注样本进行模式识别问题中的样本分布探索。针对常规雷达目标识别系统中,样本难以准确标注、模板库建立复杂以及建立过程漫长的问题,采用半监督学习算法以减少模板库的建设代价,并启用多核学习来进行目标特征的自动选择。基于窄带飞机目标分类识别的数据分析表明,与基于监督学习算法的常规识别系统相比,所提的目标识别系统能够获得更高的分类准确率。The aim of semi-supervised learning is to predict unknown labels by exploiting altogether available labeled samples and information conveyed by unlabeled data. For classical radar target recognition system that requires large-scale sample base,a narrow-band system based on semi-supervised learning is proposed to reduce the time-consuming labeling procedure,and a multi-kernel learning is employed to implement an automatic feature selection. This system is realized by the TSVM-MKL algorithm. Compared with the classical radar target recognition system based on supervised learning,the proposed system achieves better performance with the contribution of unlabeled data.

关 键 词：窄带雷达目标识别 半监督学习 多核学习 TSVM-MKL 

分 类 号：TP302.7[自动化与计算机技术—计算机系统结构]