随着社会发展，海洋空间对人类变得愈发重要，对新的水下目标自动识别系统的需求也愈发迫切。在水下目标自动识别系统的构建过程中，提取到的特征含有很多冗余特征、不相关特征和噪声特征，影响系统工作效率，降低了分类识别正确率。为此，本文提出一种新的用于水下目标识别的特征选择算法——基于图学习的无监督特征选择算法（Unsupervised Feature Selection Algorithm Based on Graph Learning，UFSGL）。该算法通过同时进行转换矩阵优化和图学习来优化算法框架，并用正则化方法优化加权图中边的光滑度，最后对转换矩阵进行稀疏化从而进行特征选择。使用UCI数据库的sonar数据集对算法性能进行验证，结果证明，UFSGL算法能够有效减少特征子集中的特征个数，并在一定程度上提高分类识别正确率。
With the development of society, marine space becomes more and more important to human beings, and the demand for new automatic identification system for underwater targets is becoming more and more urgent. In the construction of the underwater target automatic identification system, the extracted features contain many redundant, irrelevant and noise features, which affect the efficiency of the system and reduce the accuracy of classification and recognition. To this end, we proposed a new feature selection algorithm for underwater target recognition-Unsupervised Feature Selection Algorithm Based on Graph Learning (UFSGL). The algorithm framework is optimized the transformation matrix and graph learning at the same time, and use the regularization method to optimize the smoothness of the weighted edge. Using the sonar dataset of UCI database to verify the performance of the algorithm, the results show that UFSGL algorithm can effectively reduce the number of features in feature subsets and improve the accuracy of classification recognition to a certain extent.
2017,(): 91-94 收稿日期：2017-06-30
DOI：10.3404/j.issn.1672-7649.2017.12.019
分类号：TP391.4
基金项目：中国船舶系统工程研究院水声对抗国防科技重点实验室基金资助项目
作者简介：杨宏晖(1971-),女,副教授,主要从事声信号处理及模式识别的相关研究
参考文献：
[1] ROBNIKŠIKONJA M, KONONENKO I. Theoretical and empirical analysis of reliefF and RReliefF[J]. Machine Learning, 2003, 53(1-2): 23-69.
[2] DUDA R O, HART P E, STORK D G. Pattern classification. 2nd ed[J]. Pattern Analysis & Applications, 2001, 1(1): 142-143.
[3] PENG H, LONG F, DING C. Feature selection based on mutual information criteria of max-dependency, max-relevance, and min-redundancy[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2005, 27(8): 1226.
[4] MITRA P, MURTHY C A, PAL S K. Unsupervised feature selection using feature similarity[J]. Pattern Analysis & Machine Intelligence IEEE Transactions on, 2002, 24(3): 301-312.
[5] DY J G, BRODLEY C E. Feature selection for unsupervised learning[J]. Journal of Machine Learning Research, 2004, 5(4): 845-889.
[6] 申昇, 杨宏晖, 袁帅. 用于水声目标识别的互信息无监督特征选择[J]. 声学技术, 2013, 32(6): 30-33.
SHEN Sheng, YANG Hong-hui, YUAN Shuai. Unsupervised feature selection approach based on mutual information for underwater acoustic target classification[J]. Technical Acoustics, 2013, 32(6): 30-33.
[7] HE X, CAI D, NIYOGI P. Laplacian score for feature selection[C]//International Conference on Neural Information Processing Systems. MIT Press, 2005: 507-514.
[8] ZHAO Z, LIU H. Spectral feature selection for supervised and unsupervised learning[C]//Machine Learning, Proceedings of the Twenty-Fourth International Conference. DBLP, 2007: 1151-1157.
[9] CAI D, ZHANG C, HE X. Unsupervised feature selection for multi-cluster data[C]//ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. ACM, 2010: 333-342.
[10] HOU C, NIE F, LI X, et al. Joint embedding learning and sparse regression: a framework for unsupervised feature selection. [J]. IEEE Transactions on Cybernetics, 2014, 44(6): 793.
[11] X NIYOGI, “Locality preserving projections,” in Neural information processing systems. MIT, 2004.
