Abstract
Approximate k nearest neighbor (AkNN) search is a primitive operator for many applications, such as computer vision and machine learning. As these applications deal with a large set of high-dimensional points, a fast and accurate solution is required. It is known that graph-based AkNN search algorithms are faster and more accurate than other approaches, including hash- and quantization-based ones. However, existing graph-based AkNN search algorithms rely purely on heuristics, i.e., their performances are not theoretically supported. This paper proposes LGTM, a new algorithm for AkNN search, that exploits both locality-sensitive hashing and a proximity graph. The performance of LGTM is theoretically supported. Our experiments on real datasets show that LGTM outperforms state-of-the-art AkNN search algorithms.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
Notes
- 1.
In our implementation, we maintain at most 50 points in each bucket.
- 2.
- 3.
- 4.
- 5.
NSG and HNSW can process different queries in parallel, but they cannot process a query in parallel.
References
Babenko, A., Lempitsky, V.: Aggregating local deep features for image retrieval. In: ICCV, pp. 1269–1277 (2015)
Babenko, A., Lempitsky, V.: Efficient indexing of billion-scale datasets of deep descriptors. In: CVPR, pp. 2055–2063 (2016)
Bentley, J.L.: Multidimensional binary search trees used for associative searching. Commun. ACM 18(9), 509–517 (1975)
Datar, M., Immorlica, N., Indyk, P., Mirrokni, V.S.: Locality-sensitive hashing scheme based on p-stable distributions. In: SoCG, pp. 253–262 (2004)
Dong, W., Moses, C., Li, K.: Efficient k-nearest neighbor graph construction for generic similarity measures. In: WWW, pp. 577–586 (2011)
Fu, C., Xiang, C., Wang, C., Cai, D.: Fast approximate nearest neighbor search with the navigating spreading-out graph. PVLDB 12(5), 461–474 (2019)
Harwood, B., Drummond, T.: FANNG: fast approximate nearest neighbour graphs. In: CVPR, pp. 5713–5722 (2016)
Jegou, H., Douze, M., Schmid, C.: Product quantization for nearest neighbor search. IEEE Trans. Pattern Anal. Mach. Intell. 33(1), 117–128 (2010)
Li, W., et al.: Approximate nearest neighbor search on high dimensional data-experiments, analyses, and improvement. IEEE Trans. Knowl. Data Eng. 32(8), 1475–1488 (2020)
Lowe, D.G.: Distinctive image features from scale-invariant keypoints. Int. J. Comput. Vision 60(2), 91–110 (2004)
Malkov, Y.A., Yashunin, D.: Efficient and robust approximate nearest neighbor search using hierarchical navigable small world graphs. IEEE Trans. Pattern Anal. Mach. Intell. 42(4), 824–836 (2020)
Weber, R., Schek, H.J., Blott, S.: A quantitative analysis and performance study for similarity-search methods in high-dimensional spaces. In: VLDB, pp. 194–205 (1998)
Acknowledgments
This research is partially supported by JST PRESTO Grant Number JPMJPR1931, JSPS Grant-in-Aid for Scientific Research (A) Grant Number 18H04095, and JST CREST Grant Number JPMJCR21F2.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2021 Springer Nature Switzerland AG
About this paper
Cite this paper
Arai, Y., Amagata, D., Fujita, S., Hara, T. (2021). LGTM: A Fast and Accurate kNN Search Algorithm in High-Dimensional Spaces. In: Strauss, C., Kotsis, G., Tjoa, A.M., Khalil, I. (eds) Database and Expert Systems Applications. DEXA 2021. Lecture Notes in Computer Science(), vol 12924. Springer, Cham. https://doi.org/10.1007/978-3-030-86475-0_22
Download citation
DOI: https://doi.org/10.1007/978-3-030-86475-0_22
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-86474-3
Online ISBN: 978-3-030-86475-0
eBook Packages: Computer ScienceComputer Science (R0)