A self-immunizing manifold ranking for image retrieval

Jun Wu, Yidong Li, Songhe Feng, Hong Shen

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

5 Citations (Scopus)


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.

Original languageEnglish
Title of host publicationAdvances in Knowledge Discovery and Data Mining - 17th Pacific-Asia Conference, PAKDD 2013, Proceedings
Number of pages11
EditionPART 2
Publication statusPublished - 2013
Externally publishedYes
Event17th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2013 - Gold Coast, QLD, Australia
Duration: 14 Apr 201317 Apr 2013

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 2
Volume7819 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference17th Pacific-Asia Conference on Knowledge Discovery and Data Mining, PAKDD 2013
CityGold Coast, QLD


  • Content-based image retrieval
  • Elastic kNN graph
  • Local scaling
  • Relevance feedback
  • Self-immunizing manifold ranking


Dive into the research topics of 'A self-immunizing manifold ranking for image retrieval'. Together they form a unique fingerprint.

Cite this