Local Deep Learning Quantization for Approximate Nearest Neighbor Search

被引:0
作者
Li, Quan [1 ]
Xie, Xike [2 ]
Wang, Chao [1 ]
Weng, Jiali [3 ]
机构
[1] Univ Sci & Technol China, Sch Comp Sci & Technol, Hefei, Anhui, Peoples R China
[2] Univ Sci & Technol China, Suzhou Inst Adv Res, Suzhou, Jiangsu, Peoples R China
[3] Univ Sci & Technol China, Sch Software Engn USTC, Suzhou, Jiangsu, Peoples R China
来源
PROCEEDINGS OF THE 4TH ANNUAL ACM INTERNATIONAL CONFERENCE ON MULTIMEDIA RETRIEVAL, ICMR 2024 | 2024年
基金
中国国家自然科学基金;
关键词
optimal ranking; local range set; product quantization; deep learning; PRODUCT QUANTIZATION;
D O I
10.1145/3652583.3657615
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 an effective vector quantization approach to compactly encode high-dimensional vectors for approximate nearest neighbor search (ANNS). While the PQ-based ANNS methods achieve remarkable time and space efficiency, their search accuracy falls short. The main reason is the excessive quantization error of vectors used for similarity computation during the search phase. We refer to the set of these computed vectors as the local range set. We observe that if the vectors in the local range set are optimally ranked, the search accuracy will be significantly improved. Based on this observation, we propose a Local Deep Learning Quantization (LDLQ) framework. This framework involves mapping codewords to fake vectors within the local range set and utilizes fake vectors for ranking. Experimental results demonstrate that the LDLQ framework significantly improves the accuracy of existing PQ-based ANNS methods while maintaining low computation and space overhead. Notably, our method can be plugged into existing PQ-based approaches for performance enhancement, making it versatile and widely deployable.
引用
收藏
页码:1125 / 1129
页数:5
相关论文
共 33 条
[1]  
[Anonymous], 2004, Advances in neural information processing systems
[2]  
Arjovsky M, 2017, PR MACH LEARN RES, V70
[3]   ANN-Benchmarks: A benchmarking tool for approximate nearest neighbor algorithms [J].
Aumuller, Martin ;
Bernhardsson, Erik ;
Faithfull, Alexander .
INFORMATION SYSTEMS, 2020, 87
[4]   Additive Quantization for Extreme Vector Compression [J].
Babenko, Artem ;
Lempitsky, Victor .
2014 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2014, :931-938
[5]   The Inverted Multi-Index [J].
Babenko, Artem ;
Lempitsky, Victor .
2012 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2012, :3069-3076
[6]   Searching in high-dimensional spaces -: Index structures for improving the performance of multimedia Databases [J].
Böhm, C ;
Berchtold, S ;
Keim, D .
ACM COMPUTING SURVEYS, 2001, 33 (03) :322-373
[7]   Approximate Nearest Neighbor Search on Standard Search Engines [J].
Carrara, Fabio ;
Vadicamo, Lucia ;
Gennaro, Claudio ;
Amato, Giuseppe .
SIMILARITY SEARCH AND APPLICATIONS (SISAP 2022), 2022, 13590 :214-221
[8]   Con tent-based music information retrieval: Current directions and future challenges [J].
Casey, Michael A. ;
Veltkamp, Remco ;
Goto, Masataka ;
Leman, Marc ;
Rhodes, Christophe ;
Slaney, Malcolm .
PROCEEDINGS OF THE IEEE, 2008, 96 (04) :668-696
[9]   Tensor Quantization: High-Dimensional Data Compression [J].
Chang, Shih Yu ;
Wu, Hsiao-Chun .
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY, 2022, 32 (08) :5566-5580
[10]   Approximate Nearest Neighbor Search by Residual Vector Quantization [J].
Chen, Yongjian ;
Guan, Tao ;
Wang, Cheng .
SENSORS, 2010, 10 (12) :11259-11273