Codebook-softened product quantization for high accuracy approximate nearest neighbor search

被引:4
作者
Fan, Jingya [1 ]
Pan, Zhibin [1 ,3 ,4 ]
Wang, Liangzhuang [1 ]
Wang, Yang [2 ]
机构
[1] Xi An Jiao Tong Univ, Fac Elect & Informat Engn, Xian 710049, Shaanxi, Peoples R China
[2] Xian Univ Technol, Dept Informat Sci, Xian 710048, Shaanxi, Peoples R China
[3] Zhengzhou Xinda Inst Adv Technol, Zhengzhou 450002, Peoples R China
[4] Xi An Jiao Tong Univ, Res Inst, Hangzhou 311215, Zhejiang, Peoples R China
基金
中国国家自然科学基金;
关键词
Approximate nearest neighbor search; Product quantization; Vector quantization; High-dimensional space; Soften codebooks; BINARY QUANTIZATION; NORMALITY; CODES;
D O I
10.1016/j.neucom.2022.08.002
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Product quantization (PQ) is a fundamental technique for approximate nearest neighbor (ANN) search in many applications such as information retrieval, computer vision and pattern recognition. In the existing PQ-based methods for approximate nearest neighbor search, the reachable best search accuracy is achieved by using fixed codebooks. The search performance is limited by the quality of the hard codebooks. Unlike the existing methods, in this paper, we present a novel codebook-softened product quantization (CSPQ) method to achieve more quantization levels by softening codebooks. We firstly analyze how well the database vectors match the trained codebooks by examining quantization error for each database vector, and select the bad-matching database vectors. Then, we give the trained codebooks bbit freedom to soften codebooks. Finally, by minimizing quantization errors, the bad-matching vectors are encoded by softened codebooks and the labels of best-matching codebooks are recorded. Experimental results on SIFT1M, GIST1M and SIFT10M show that, despite its simplicity, our proposed method achieves higher accuracy compared with PQ and it can be combined with other nonexhaustive frameworks to achieve fast search. (c) 2022 Elsevier B.V. All rights reserved.
引用
收藏
页码:107 / 116
页数:10
相关论文
共 50 条
[21]   Optimized residual vector quantization for efficient approximate nearest neighbor search [J].
Liefu Ai ;
Junqing Yu ;
Zebin Wu ;
Yunfeng He ;
Tao Guan .
Multimedia Systems, 2017, 23 :169-181
[22]   Distance Encoded Product Quantization for Approximate K-Nearest Neighbor Search in High-Dimensional Space [J].
Heo, Jae-Pil ;
Lin, Zhe ;
Yoon, Sung-Eui .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2019, 41 (09) :2084-2097
[23]   Local Deep Learning Quantization for Approximate Nearest Neighbor Search [J].
Li, Quan ;
Xie, Xike ;
Wang, Chao ;
Weng, Jiali .
PROCEEDINGS OF THE 4TH ANNUAL ACM INTERNATIONAL CONFERENCE ON MULTIMEDIA RETRIEVAL, ICMR 2024, 2024, :1125-1129
[24]   Effective product quantization-based indexing for nearest neighbor search [J].
Chiu, Chih-Yi ;
Chiu, Jih-Sheng ;
Markchit, Sarawut ;
Chou, Sheng-Hao .
MULTIMEDIA TOOLS AND APPLICATIONS, 2019, 78 (03) :2877-2895
[25]   Effective product quantization-based indexing for nearest neighbor search [J].
Chih-Yi Chiu ;
Jih-Sheng Chiu ;
Sarawut Markchit ;
Sheng-Hao Chou .
Multimedia Tools and Applications, 2019, 78 :2877-2895
[26]   Iteratively Multiple Projections Optimization for Product Quantization in Nearest Neighbor Search [J].
Li, Jin ;
Lan, Xuguang ;
Zheng, Nanning .
2017 IEEE INTERNATIONAL CONFERENCE ON BIG KNOWLEDGE (IEEE ICBK 2017), 2017, :65-71
[27]   STACKED PRODUCT QUANTIZATION FOR NEAREST NEIGHBOR SEARCH ON LARGE DATASETS [J].
Wang, Jun ;
Li, Zhiyang ;
Du, Yegang ;
Qu, Wenyu .
2016 IEEE TRUSTCOM/BIGDATASE/ISPA, 2016, :1621-1627
[28]   Accelerated Approximate Nearest Neighbors Search Through Hierarchical Product Quantization [J].
Abdelhadi, Ameer M. S. ;
Bouganis, Christos-Savvas ;
Constantinides, George A. .
2019 INTERNATIONAL CONFERENCE ON FIELD-PROGRAMMABLE TECHNOLOGY (ICFPT 2019), 2019, :90-98
[29]   Approximate Nearest Neighbor Search on High Dimensional Data - Experiments, Analyses, and Improvement [J].
Li, Wen ;
Zhang, Ying ;
Sun, Yifang ;
Wang, Wei ;
Li, Mingjie ;
Zhang, Wenjie ;
Lin, Xuemin .
IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, 2020, 32 (08) :1475-1488
[30]   Quantization-Based Approximate Nearest Neighbor Search with Optimized Multiple Residual Codebooks [J].
Uchida, Yusuke ;
Takagi, Koichi ;
Kawada, Ryoichi .
IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS, 2011, E94D (07) :1510-1514