Supervised Ranking Hash For Semantic Similarity Search
Keywords
Image retrieval; Rank correlation; Similarity search; Subspace learning; Supervised Hashing
Abstract
The era of big data has spawned unprecedented interests in developing hashing algorithms for their storage efficiency and effectiveness in fast nearest neighbor search in large-scale databases. Most of the existing hash learning algorithms focus on learning hash functions which generate binary codes by numeric quantization of some projected feature space. In this work, we propose a novel hash learning framework that encodes features' ranking orders instead of quantizing their numeric values in a number of optimal lowdimensional ranking subspaces. We formulate the rankingbased hash learning problem as the optimization of a continuous probabilistic error function using softmax approximation and present an efficient learning algorithm to solve the problem. Our work is a generalization of the Winner-Take-All (WTA) hashing algorithm and naturally enjoys the numeric stability benefits of rank correlation measures while being optimized to achieve high precision at extremely short code length. We extensively evaluate the proposed algorithm in several datasets and demonstrate superior performance against several state-of-the-arts.
Publication Date
1-18-2017
Publication Title
Proceedings - 2016 IEEE International Symposium on Multimedia, ISM 2016
Number of Pages
551-558
Document Type
Article; Proceedings Paper
Personal Identifier
scopus
DOI Link
https://doi.org/10.1109/ISM.2016.133
Copyright Status
Unknown
Socpus ID
85015162292 (Scopus)
Source API URL
https://api.elsevier.com/content/abstract/scopus_id/85015162292
STARS Citation
Li, Kai; Qi, Guo Jun; Ye, Jun; Yusuph, Tuoerhongjiang; and Hua, Kien A., "Supervised Ranking Hash For Semantic Similarity Search" (2017). Scopus Export 2015-2019. 7449.
https://stars.library.ucf.edu/scopus2015/7449