Optimized K-Means Hashing for Approximate Nearest Neighbor Search

Abstract:

Hashing which maps data into binary codes in Hamming space has attracted more and more attentions for approximate nearest neighbor search due to its high efficiency and reduced storage cost. K-means hashing (KH) is a novel hashing method which firstly quantizes the data by codewords and then uses the indices of codewords to encode the data. However, in KH, only the codewords are updated to minimize the quantization error and affinity error while the indices of codewords remain the same after they are initialized. In this paper, we propose an optimized k-means hashing (OKH) method to encode data by binary codes. In our method, we simultaneously optimize the codewords and the indices of them to minimize the quantization error and the affinity error. Our OKH method can find both the optimal codewords and the optiaml indices, and the resulting binary codes in Hamming space can better preserve the original neighborhood structure of the data. Besides, OKH can further be generalized to a product space. Extensive experiments have verified the superiority of OKH over KH and other state-of-the-art hashing methods.

• Periodical: Applied Mechanics and Materials (Volumes 651-653)
• Pages: 2168-2171
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.651-653.2168
• Online since: September 2014
• Authors: Qin Zhen Guo*, Zhi Zeng, Shu Wu Zhang, Yuan Zhang, Gui Xuan Zhang
• Keywords: Approximate Nearest Neighbor Search, Hashing, Image Retrieval, Vector Quantization

References

[1] J. Beis, and D. Lowe, in: CVPR, 1997, p.1000.

[2] H. Jegou, M. Douze, and C. Schmid, IEEE TPAMI, Vol. 33 (2011) p.117.

[3] Q. -Z. Guo, Z. Zeng, S. Zhang, Y. Zhang, and F. Wang, in: ICME, 2013, p.1.

[4] K. He, F. Wen, J. Sun, in: CVPR, 2013, p.2938.

[5] B. Wang, Z. Li and M. Li, in: ICME, 2006, p.353.

[6] M. Datar, N. Immorlica, P. Indyk, and V. Mirrokni, in: Annu. Symp. Computational Geometry, ACM, 2004, p.253.

[7] Y. Weiss, A. Torralba and R. Fergus, in NIPS, 2008, p.1753.

[8] Y. Gong and S. Lazebnik, in CVPR, 2011, p.817.