摘要
基于统计学习理论和遗传算法理论,提出了一种基于支持向量机和遗传算法相结合的水下目标特征选择算法。通过对实测数据的特征集的优化选择实验,证明了该算法的有效性和鲁棒性,它能较好地解决在复杂水下目标信号所提取的特征维数高,样本采样困难,数目偏少的实际情况下的分类识别问题。
It is crucial to find an effective feature selection method for underwater acoustic targets. In this paper, a new method is proposed for feature selection that uses support vector machines （SVMs） combined with genetic algorithm （GA）. GA is used for selecting optimal feature subset on the basis of predicted accuracy of SVMs. The SVM-GA is compared with Sequential Back Selection （SBS）, a kind of typical and popular feature selection method. Three different classes of underwater targets datasets are used in the experiment. The results show that the SVM-GA is more effective than SBS for the purpose of feature selection for underwater targets, and with this method the dimension of training data is reduced about 50% while the classification accuracy is almost the same. We conclude that the SVM-GA is an effective feature selection technique to select feature subset of underwater acoustic targets.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
2005年第4期512-515,共4页
Journal of Northwestern Polytechnical University
关键词
特征选择
支持向量机
遗传算法
feature selection, support vector machine （SVM）, genetic algorithm （GA）
