Vector search with small radiuses

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Szilvasy, Gergely, Mazaré, Pierre-Emmanuel, Douze, Matthijs
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909138351554560
author Szilvasy, Gergely
Mazaré, Pierre-Emmanuel
Douze, Matthijs
author_facet Szilvasy, Gergely
Mazaré, Pierre-Emmanuel
Douze, Matthijs
contents In recent years, the dominant accuracy metric for vector search is the recall of a result list of fixed size (top-k retrieval), considering as ground truth the exact vector retrieval results. Although convenient to compute, this metric is distantly related to the end-to-end accuracy of a full system that integrates vector search. In this paper we focus on the common case where a hard decision needs to be taken depending on the vector retrieval results, for example, deciding whether a query image matches a database image or not. We solve this as a range search task, where all vectors within a certain radius from the query are returned. We show that the value of a range search result can be modeled rigorously based on the query-to-vector distance. This yields a metric for range search, RSM, that is both principled and easy to compute without running an end-to-end evaluation. We apply this metric to the case of image retrieval. We show that indexing methods that are adapted for top-k retrieval do not necessarily maximize the RSM. In particular, for inverted file based indexes, we show that visiting a limited set of clusters and encoding vectors compactly yields near optimal results.
format Preprint
id arxiv_https___arxiv_org_abs_2403_10746
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Vector search with small radiuses
Szilvasy, Gergely
Mazaré, Pierre-Emmanuel
Douze, Matthijs
Computer Vision and Pattern Recognition
Databases
In recent years, the dominant accuracy metric for vector search is the recall of a result list of fixed size (top-k retrieval), considering as ground truth the exact vector retrieval results. Although convenient to compute, this metric is distantly related to the end-to-end accuracy of a full system that integrates vector search. In this paper we focus on the common case where a hard decision needs to be taken depending on the vector retrieval results, for example, deciding whether a query image matches a database image or not. We solve this as a range search task, where all vectors within a certain radius from the query are returned. We show that the value of a range search result can be modeled rigorously based on the query-to-vector distance. This yields a metric for range search, RSM, that is both principled and easy to compute without running an end-to-end evaluation. We apply this metric to the case of image retrieval. We show that indexing methods that are adapted for top-k retrieval do not necessarily maximize the RSM. In particular, for inverted file based indexes, we show that visiting a limited set of clusters and encoding vectors compactly yields near optimal results.
title Vector search with small radiuses
topic Computer Vision and Pattern Recognition
Databases
url https://arxiv.org/abs/2403.10746