基于抗噪声邻域粗糙集的在线流特征选择算法

Online Streaming Feature Selection Based on Anti-noise Neighborhood Rough Set

在开放动态环境当中,特征是动态生成的,特征在不同时间戳内流入特征空间称为流特征.然而,在一些基于传统的邻域粗糙集流特征选择算法中,噪声点会对特征的依赖度计算造成影响.基于此,本文提出了基于抗噪声邻域粗糙集的在线流特征选择算法.首先,充分考虑噪声点的影响,定义一种抗噪声的邻域关系,并设计基于抗噪声邻域的依赖度计算公式.进一步,考虑到特征对不同类别所提供的信息不同,结合类别正域,提出了一种新的在线相关性分析方法和冗余分析方法.在8个数据集上的实验研究表明,所提算法得到的特征子集优于一些在线流特征选择算法.

In an open and dynamic environment,featrures are dynamically generated,and feature flows into the feature space in different time stamps is called streaming feature.However,in some traditional online streaming feature selection algorithms based on neighborhood rough sets,the noise samples will affect the feature dependence.To address this problem,an online feature selection algorithm based on anti-noise neighborhood rough set is proposed.First,an anti-noise neighborhood is defined via considering the influence of noise sample.Then,the formula of dependency with respect to anti-noise neighborhood is designded.Forthermore,we propose a new online relevence analysis and redundant analysis for selecting useful features,which can provide relevant information for different class.By shown in experimental studies on eight datasets,the proposed algorithm outperforms some state-of-the-art algorithms.