Competitive Quantization for Approximate Nearest Neighbor Search

被引:21
作者
Ozan, Ezgi Can [1 ]
Kiranyaz, Serkan [2 ]
Gabbouj, Moncef [1 ]
机构
[1] Tampere Univ Technol, Dept Signal Proc, Tampere 33720, Finland
[2] Qatar Univ, Coll Engn, Dept Elect Engn, Doha 2713, Qatar
关键词
Approximate nearest neighbor search; binary codes; large-scale retrieval; vector quantization;
D O I
10.1109/TKDE.2016.2597834
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In this study, we propose a novel vector quantization algorithm for Approximate Nearest Neighbor (ANN) search, based on a joint competitive learning strategy and hence called as competitive quantization (CompQ). CompQ is a hierarchical algorithm, which iteratively minimizes the quantization error by jointly optimizing the codebooks in each layer, using a gradient decent approach. An extensive set of experimental results and comparative evaluations show that CompQ outperforms the-state-of-the-art while retaining a comparable computational complexity.
引用
收藏
页码:2884 / 2894
页数:11
相关论文
共 29 条
  • [1] Ai L., 2015, PROC MULTIMEDIA SYST, P1
  • [2] [Anonymous], 2004, P 20 ACM S COMP
  • [3] [Anonymous], 2009, NEURIPS, P1753
  • [4] [Anonymous], 2014, ARXIV14082927
  • [5] [Anonymous], P 31 INT C INT C MAC
  • [6] [Anonymous], 2008, Proceedings of the 31st Annual International ACM SIGIR Conference on Research and Development in Information Retrieval
  • [7] Babenko A, 2015, PROC CVPR IEEE, P4240, DOI 10.1109/CVPR.2015.7299052
  • [8] Additive Quantization for Extreme Vector Compression
    Babenko, Artem
    Lempitsky, Victor
    [J]. 2014 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2014, : 931 - 938
  • [9] The Inverted Multi-Index
    Babenko, Artem
    Lempitsky, Victor
    [J]. 2012 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2012, : 3069 - 3076
  • [10] Transform Coding for Fast Approximate Nearest Neighbor Search in High Dimensions
    Brandt, Jonathan
    [J]. 2010 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2010, : 1815 - 1822