學術產出-Proceedings
Article View/Open
Publication Export
-
題名 Network based discriminant analysis for multiclassification 作者 陳立榜
Chen, Li-Pang貢獻者 統計系 日期 2021-12 上傳時間 7-Oct-2022 13:43:11 (UTC+8) 摘要 With the rapid advance of information technology, complex data are collected easily, and undoubtedly, they become more challenging than we expect. In addition, network structure is important feature and is ubiquitous in high-dimensional data because of strong or weak correlations among variables. Our main interest is to use predictors to do multiclassification. While discriminant analysis is one of supervised learning methods to deal with multiclassification and relevant extensions have been explored, little method has been available to deal with multiclassification with network structures accommodated. To incorporate the network structure and improve the accuracy of classification, we propose network based linear discriminant analysis and quadratic discriminant analysis in this paper. The main advantage of the proposed methods is to incorporate network structure of predictors and analyze the classification with multiclass responses instead of restricting on binary responses. In addition, the proposed methods are easy to compute and implement. Finally, numerical studies are conducted to assess the performance of the proposed methods, and numerical results verify that the proposed methods outperform their competitors. 關聯 2021年統計學術研討會暨台、日、韓國際統計學術研討會, 輔仁大學統計資訊學系、中國統計學社 資料類型 conference dc.contributor 統計系 dc.creator (作者) 陳立榜 dc.creator (作者) Chen, Li-Pang dc.date (日期) 2021-12 dc.date.accessioned 7-Oct-2022 13:43:11 (UTC+8) - dc.date.available 7-Oct-2022 13:43:11 (UTC+8) - dc.date.issued (上傳時間) 7-Oct-2022 13:43:11 (UTC+8) - dc.identifier.uri (URI) http://nccur.lib.nccu.edu.tw/handle/140.119/142350 - dc.description.abstract (摘要) With the rapid advance of information technology, complex data are collected easily, and undoubtedly, they become more challenging than we expect. In addition, network structure is important feature and is ubiquitous in high-dimensional data because of strong or weak correlations among variables. Our main interest is to use predictors to do multiclassification. While discriminant analysis is one of supervised learning methods to deal with multiclassification and relevant extensions have been explored, little method has been available to deal with multiclassification with network structures accommodated. To incorporate the network structure and improve the accuracy of classification, we propose network based linear discriminant analysis and quadratic discriminant analysis in this paper. The main advantage of the proposed methods is to incorporate network structure of predictors and analyze the classification with multiclass responses instead of restricting on binary responses. In addition, the proposed methods are easy to compute and implement. Finally, numerical studies are conducted to assess the performance of the proposed methods, and numerical results verify that the proposed methods outperform their competitors. dc.format.extent 2004663 bytes - dc.format.mimetype application/pdf - dc.relation (關聯) 2021年統計學術研討會暨台、日、韓國際統計學術研討會, 輔仁大學統計資訊學系、中國統計學社 dc.title (題名) Network based discriminant analysis for multiclassification dc.type (資料類型) conference
