
For FullText PDF, please login, if you are a member of IEICE,
or go to Pay Per View on menu list, if you are a nonmember of IEICE.

Partial Label Metric Learning Based on Statistical Inference
Tian XIE Hongchang CHEN Tuosiyu MING Jianpeng ZHANG Chao GAO Shaomei LI Yuehang DING
Publication
IEICE TRANSACTIONS on Information and Systems
Vol.E103D
No.6
pp.13551361 Publication Date: 2020/06/01
Online ISSN: 17451361
DOI: 10.1587/transinf.2019EDP7182
Type of Manuscript: PAPER Category: Artificial Intelligence, Data Mining Keyword: partial label learning, metric learning, statistical inference, likelihoodratio test,
Full Text: PDF(1.5MB)>>
Summary:
In partial label data, the groundtruth label of a training example is concealed in a set of candidate labels associated with the instance. As the groundtruth label is inaccessible, it is difficult to train the classifier via the label information. Consequently, manifold structure information is adopted, which is under the assumption that neighbor/similar instances in the feature space have similar labels in the label space. However, the realworld data may not fully satisfy this assumption. In this paper, a partial label metric learning method based on likelihoodratio test is proposed to make partial label data satisfy the manifold assumption. Moreover, the proposed method needs no objective function and treats the data pairs asymmetrically. The experimental results on several realworld PLL datasets indicate that the proposed method outperforms the existing partial label metric learning methods in terms of classification accuracy and disambiguation accuracy while costs less time.

