Multi-class classification via discriminative multiple subspace learning
Tang, Tang; Qiao, Hong; Zheng, Suiwu; Tang, T
2013
Conference NameInternational Conference on Mechatronic Sciences, Electric Engineering and Computer (MEC)
Source PublicationPROCEEDINGS 2013 INTERNATIONAL CONFERENCE ON MECHATRONIC SCIENCES, ELECTRIC ENGINEERING AND COMPUTER (MEC)
Conference DateDEC 20-22, 2013
Conference PlaceShenyang, PEOPLES R CHINA
AbstractSubspace learning has long been a fundamental yet important problem of modeling data distributions. In this paper, we propose to learn multiple linear subspaces in a supervised way for multi-classclassification. To this end, a discriminative term redefining decision margin in terms of reconstruction error is incorporated into the model. The term enjoys similar properties of hinge loss function to the benefit of classification and leads to a training process seeking the balance between unsupervisedlearning and supervised learning. In the experiments on written digits dataset, our algorithm outperforms other methods proposed recently in both accuracy and computation efficiency.
KeywordSubspace Learning Generative Model Discriminative Model
Document Type会议论文
Identifierhttp://ir.ia.ac.cn/handle/173211/12868
Collection复杂系统管理与控制国家重点实验室_机器人理论与应用
Corresponding AuthorTang, T
Recommended Citation
GB/T 7714
Tang, Tang,Qiao, Hong,Zheng, Suiwu,et al. Multi-class classification via discriminative multiple subspace learning[C],2013.
Files in This Item:
There are no files associated with this item.
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Tang, Tang]'s Articles
[Qiao, Hong]'s Articles
[Zheng, Suiwu]'s Articles
Baidu academic
Similar articles in Baidu academic
[Tang, Tang]'s Articles
[Qiao, Hong]'s Articles
[Zheng, Suiwu]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Tang, Tang]'s Articles
[Qiao, Hong]'s Articles
[Zheng, Suiwu]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.