Projection Bank: From High-dimensional Data to Medium-length Binary Codes

被引:14
|
作者
Liu, Li [1 ]
Yu, Mengyang [1 ]
Shao, Ling [1 ]
机构
[1] Northumbria Univ, Dept Comp Sci & Digital Technol, Newcastle Upon Tyne NE1 8ST, Tyne & Wear, England
来源
2015 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV) | 2015年
关键词
D O I
10.1109/ICCV.2015.323
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Recently, very high-dimensional feature representations, e.g., Fisher Vector, have achieved excellent performance for visual recognition and retrieval. However, these lengthy representations always cause extremely heavy computational and storage costs and even become unfeasible in some large-scale applications. A few existing techniques can transfer very high-dimensional data into binary codes, but they still require the reduced code length to be relatively long to maintain acceptable accuracies. To target a better balance between computational efficiency and accuracies, in this paper, we propose a novel embedding method called Binary Projection Bank (BPB), which can effectively reduce the very high-dimensional representations to medium-dimensional binary codes without sacrificing accuracies. Instead of using conventional single linear or bilinear projections, the proposed method learns a bank of small projections via the max-margin constraint to optimally preserve the intrinsic data similarity. We have systematically evaluated the proposed method on three datasets: Flickr 1M, ILSVR2010 and UCF101, showing competitive retrieval and recognition accuracies compared with state-of-the-art approaches, but with a significantly smaller memory footprint and lower coding complexity.
引用
收藏
页码:2821 / 2829
页数:9
相关论文
共 50 条
  • [31] DPPro: Differentially Private High-Dimensional Data Release via Random Projection
    Xu, Chugui
    Ren, Ju
    Zhang, Yaoxue
    Qin, Zhan
    Ren, Kui
    IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY, 2017, 12 (12) : 3081 - 3093
  • [32] Distance-preserving projection of high-dimensional data for nonlinear dimensionality reduction
    Yang, L
    IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2004, 26 (09) : 1243 - 1246
  • [33] Voronoi Constellations for High-Dimensional Lattice Codes
    Ferdinand, Nuwan S.
    Noklehy, Matthew
    Aazhang, Behnaam
    2016 IEEE INTERNATIONAL SYMPOSIUM ON INFORMATION THEORY, 2016, : 1506 - 1510
  • [34] The construction of non-coherent space time codes from high-dimensional spherical codes
    Utkovski, Zoran
    Lindner, Juergen
    2006 IEEE NINTH INTERNATIONAL SYMPOSIUM ON SPREAD SPECTRUM TECHNIQUES AND APPLICATIONS, PROCEEDINGS, 2006, : 327 - 331
  • [35] Low-Dimensional Shaping for High-Dimensional Lattice Codes
    Ferdinand, Nuwan S.
    Kurkoski, Brian M.
    Nokleby, Matthew
    Aazhang, Behnaam
    IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, 2016, 15 (11) : 7405 - 7418
  • [36] On Criticality in High-Dimensional Data
    Saremi, Saeed
    Sejnowski, Terrence J.
    NEURAL COMPUTATION, 2014, 26 (07) : 1329 - 1339
  • [37] High-Dimensional Data Bootstrap
    Chernozhukov, Victor
    Chetverikov, Denis
    Kato, Kengo
    Koike, Yuta
    ANNUAL REVIEW OF STATISTICS AND ITS APPLICATION, 2023, 10 : 427 - 449
  • [38] High-dimensional data clustering
    Bouveyron, C.
    Girard, S.
    Schmid, C.
    COMPUTATIONAL STATISTICS & DATA ANALYSIS, 2007, 52 (01) : 502 - 519
  • [40] A joint normal-binary (probit) model for high-dimensional longitudinal data
    Delporte, Margaux
    Fieuws, Steffen
    Molenberghs, Geert
    Verbeke, Geert
    De Coninck, David
    Hoorens, Vera
    STATISTICAL MODELLING, 2025, 25 (01) : 13 - 34