A graph optimization method for dimensionality reduction with pairwise constraints

详细信息    查看全文

• 作者：Limei Zhang ; Lishan Qiao
• 关键词：Dimensionality reduction ; Semi ; supervised learning ; Graph construction ; Pairwise constraints ; Locality preserving projections
• 刊名：International Journal of Machine Learning and Cybernetics
• 出版年：2017
• 出版时间：February 2017
• 年：2017
• 卷：8
• 期：1
• 页码：275-281
• 全文大小：
• 刊物类别：Engineering
• 刊物主题：Computational Intelligence; Artificial Intelligence (incl. Robotics); Control, Robotics, Mechatronics; Complex Systems; Systems Biology; Pattern Recognition;
• 出版者：Springer Berlin Heidelberg
• ISSN：1868-808X
• 卷排序：8

文摘

Graph is at the heart of many dimensionality reduction (DR) methods. Despite its importance, how to establish a high-quality graph is currently a pursued problem. Recently, a new DR algorithm called graph-optimized locality preserving projections (GoLPP) was proposed to perform graph construction with DR simultaneously in a unified objective function, resulting in an automatically optimized graph rather than pre-specified one as involved in typical LPP. However, GoLPP is unsupervised and can not naturally incorporate supervised information due to a strong sum-to-one constraint of weights of graph in its model. To address this problem, in this paper we give an improved GoLPP model by relaxing the constraint, and then develop a semi-supervised GoLPP (S-GoLPP) algorithm by incorporating pairwise constraint information into its modeling. Interestingly, we obtain a semi-supervised closed-form graph-updating formulation with natural possibility explanation. The feasibility and effectiveness of the proposed method is verified on several publicly available UCI and face data sets.