Please use this identifier to cite or link to this item: http://hdl.handle.net/2440/88130
Citations
Scopus Web of Science® Altmetric
?
?
Type: Conference paper
Title: A self-immunizing manifold ranking for image retrieval
Author: Wu, J.
Li, Y.
Feng, S.
Shen, H.
Citation: Advances in Knowledge Discovery and Data Mining: 17th Pacific-Asia Conference, PAKDD 2013, Gold Coast, Australia, April 14-17, 2013, Proceedings, Part II, 2013 / Pei, J., Tseng, V., Cao, L., Motoda, H., Xu, G. (ed./s), vol.7819 LNAI, iss.PART 2, pp.426-436
Publisher: Springer
Publisher Place: Germany
Issue Date: 2013
Series/Report no.: Lecture Notes in Computer Science; 7819
ISBN: 9783642374555
ISSN: 0302-9743
1611-3349
Conference Name: 17th Pacific-Asia Conference, PAKDD 2013 (14 Apr 2013 - 17 Apr 2013 : Gold Coast, Australia)
Statement of
Responsibility: 
Jun Wu, Yidong Li, Songhe Feng, and Hong Shen
Abstract: Manifold ranking (MR), as a powerful semi-supervised learning algorithm, plays an important role to deal with the relevance feedback problem in content-based image retrieval (CBIR). However, conventional MR has two main drawbacks: 1) in many cases, it is prone to exploit “unreliable” unlabeled images when deployed in CBIR due to the semantic gap; 2) the performance of MR is quite sensitive to the scale parameter used for calculating the Laplacian matrix. In this work, a self-immunizing MR approach is presented to address the drawbacks. Concretely, we first propose an elastic kNN graph as well as its constructing algorithm to exploit unlabeled images “safely”, and then develop a local scaling solution to calculate the Laplacian matrix adaptively. Extensive experiments on 10,000 Corel images show that the proposed algorithm is more effective than the state-of-the-art approaches.
Keywords: content-based image retrieval; relevance feedback; self-immunizing manifold ranking; elastic kNN graph; local scaling
Rights: © Springer-Verlag Berlin Heidelberg 2013
RMID: 0020134930
DOI: 10.1007/978-3-642-37456-2_36
Appears in Collections:Computer Science publications

Files in This Item:
There are no files associated with this item.


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