多模场景下的高维数据的特征选择及分类研究  

Research on feature selection and classification for high dimensional data in multimodal scene

作　　者：邓万宇[1] 刘丹[1] 董莹莹 张莎莎 刘光达 DENG Wan-yu;LIU Dan;DONG Ying-ying;ZHANG Sha-sha;LIU Guang-da(School of Computer,Xi’ an University of Post ＆ Telecommunications,Xi’ an 710061,China)

机构地区：[1]西安邮电大学计算机院,西安710061

出　　处：《信息技术》2018年第7期39-42,共4页Information Technology

基　　金：西安邮电大学创新基金项目(CXL2016-39)

摘　　要：近来,有效和准确地诊断阿尔兹海默症已经引起了越来越多的关注。然而,许多的研究只关注于单模态,并且现有的中小规模特征选择方法很难有效地从大量的特征集合中识别出相关的特征子集。文中围绕多模数据融合,旨在研究高维中等样本的特征选择算法,能够有效识别与疾病紧密相关的特征,进而从健康对照组中准确地分类出AD患者。结果表明,文中所提出的方法分类性能优于其它方法。Recently,it is attracted more and more attention to diagnose Alzheimer＇s disease（AD）effectively and accurately. However,many research focus on only a single modality,and most existing small and medium scale feature selection method make it difficult to identify relevant feature subsets from a large number of feature sets. The method focus on multi-modality data fusion,which aims to study feature selection algorithm of high dimensional data,and it can effectively identify relevant disease feature and then successfully classify AD from healthy controls. As a result, the method shows better classification performance than other methods.

关 键 词：阿尔兹海默症 多模数据 特征选择 分类性能 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]