Characterizing the optimal pivots for efficient similarity searches in vector space databases with Minkowski distances

Characterizing the optimal pivots for efficient similarity searches in vector space databases... Pivot-based retrieval algorithms are commonly used to solve similarity queries in a number of application domains, such as multimedia retrieval, biomedical databases, time series and computer vision. The query performances of pivot-based index algorithms can be significantly improved by properly choosing the set of pivots that is able to narrow down the database elements to only those relevant to a query. While many other approaches in the literature rely on empirical studies or intuitive observations and assumptions to achieve effective pivot strategies, this paper addresses the problem by using a formal mathematical approach. We conclude in our study that the optimal set of pivots in vector databases with Lp metrics is a set of uniformly distributed points on the surface of an n-sphere defined by these metrics. To make the study mathematically tractable, a uniform distribution of data in the database is assumed, allowing us to outline the problem from a purely geometrical point of view. Then, we present experimental results demonstrating the usefulness of our characterization when applied to real databases in the (Rn,Lp) metric space. Our technique is shown to outperform comparable techniques in the literature. However, we do not propose a new pivot-selection technique but rather experiments that are designed exclusively to show the usefulness of such a characterization. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Applied Mathematics and Computation Elsevier

Characterizing the optimal pivots for efficient similarity searches in vector space databases with Minkowski distances

Loading next page...
 
/lp/elsevier/characterizing-the-optimal-pivots-for-efficient-similarity-searches-in-eEX74xhOPq
Publisher
Elsevier
Copyright
Copyright © 2018 Elsevier Inc.
ISSN
0096-3003
eISSN
1873-5649
D.O.I.
10.1016/j.amc.2018.01.028
Publisher site
See Article on Publisher Site

Abstract

Pivot-based retrieval algorithms are commonly used to solve similarity queries in a number of application domains, such as multimedia retrieval, biomedical databases, time series and computer vision. The query performances of pivot-based index algorithms can be significantly improved by properly choosing the set of pivots that is able to narrow down the database elements to only those relevant to a query. While many other approaches in the literature rely on empirical studies or intuitive observations and assumptions to achieve effective pivot strategies, this paper addresses the problem by using a formal mathematical approach. We conclude in our study that the optimal set of pivots in vector databases with Lp metrics is a set of uniformly distributed points on the surface of an n-sphere defined by these metrics. To make the study mathematically tractable, a uniform distribution of data in the database is assumed, allowing us to outline the problem from a purely geometrical point of view. Then, we present experimental results demonstrating the usefulness of our characterization when applied to real databases in the (Rn,Lp) metric space. Our technique is shown to outperform comparable techniques in the literature. However, we do not propose a new pivot-selection technique but rather experiments that are designed exclusively to show the usefulness of such a characterization.

Journal

Applied Mathematics and ComputationElsevier

Published: Jul 1, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off