基于图拉普拉斯的多标签类属特征选择  

Multi-label label-specific feature selection based on graph Laplacian

作　　者：吴喆君 黄睿[1] WU Zhejun;HUANG Rui(School of Communication and Information Engineering,Shanghai University,Shanghai 200444,China)

机构地区：[1]上海大学通信与信息工程学院,上海200444

出　　处：《上海大学学报（自然科学版）》2022年第2期281-290,共10页Journal of Shanghai University:Natural Science Edition

基　　金：国家自然科学基金资助项目(61671283)。

摘　　要：多标签特征选择能够有效去除冗余特征并提升分类精度,是解决“维数灾难”问题的有效方法.然而,已有的多标签特征选择算法是对所有标签选择出相同的特征,忽略了标签与特征之间的内在联系.事实上,每个标签都具有反映该标签特有属性的特征,即类属特征.提出一种基于图拉普拉斯的多标签类属特征选择(multi-label label-specific feature selection based on graph Laplacian,LSGL)算法.对于每个类别标签,基于拉普拉斯映射获得数据的低维嵌入,再通过稀疏正则化获得数据空间到嵌入空间的投影矩阵,接着通过分析矩阵系数确定每个标签相应的类属特征,最后使用类属特征进行分类.在5个公共多标签数据集上的多标签特征选择与分类实验结果证明了所提算法的有效性.Multi-label feature selection,which can effectively remove redundant features and improve classification performance,has become an effective solution for the problem of"curse of dimensionality".However,existing multi-label feature selection methods select the same features for all labels without considering the intrinsic relation between labels and features.In fact,each label has label-specific features that reflect the specific attributes of the label.A feature selection method called multi-label label-specific feature selection based on graph Laplacian(LSGL)is proposed in this study.LSGL first obtains a low-dimensional embedding of instances for each class label based on Laplacian eigenmaps.Next,it obtains a projection matrix that can project samples from a data space to manifold embedding space through sparse regularization.It then determines the label-specific features of the corresponding class label by coefficient analysis of the matrix.Finally,the label-specific features are used for classification.Experimental results of multi-label feature selection and classification on five public multi-label datasets showed the effectiveness of the proposed algorithm.

关 键 词：多标签学习 特征选择 类属特征 图拉普拉斯 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]