On the Application of Convex Transforms to Metric Search

J 2020

On the Application of Convex Transforms to Metric Search

CONNOR, Richard, Alan DEARLE, Vladimír MÍČ and Pavel ZEZULA

Basic information

Original name

On the Application of Convex Transforms to Metric Search

Authors

CONNOR, Richard (826 United Kingdom of Great Britain and Northern Ireland), Alan DEARLE (826 United Kingdom of Great Britain and Northern Ireland), Vladimír MÍČ (203 Czech Republic, belonging to the institution) and Pavel ZEZULA (203 Czech Republic, belonging to the institution)

Edition

Pattern Recognition Letters, 2020, 0167-8655

Other information

Language

English

Type of outcome

Článek v odborném periodiku

Field of Study

10201 Computer sciences, information science, bioinformatics

Country of publisher

Netherlands

Confidentiality degree

není předmětem státního či obchodního tajemství

References:

URL

Impact factor

Impact factor: 3.756

RIV identification code

RIV/00216224:14330/20:00116150

Organization unit

Faculty of Informatics

DOI

http://dx.doi.org/10.1016/j.patrec.2020.08.008

UT WoS

000579804900076

Keywords in English

similarity search; transformation of distance function; metric space; convex transform

Abstract

V originále

Scalable similarity search in metric spaces relies on using the mathematical properties of the space in order to allow efficient querying. Most important in this context is the triangle inequality property, which can allow the majority of individual similarity comparisons to be avoided for a given query. However many important metric spaces, typically those with high dimensionality, are not amenable to such techniques. In the past convex transforms have been studied as a pragmatic mechanism which can overcome this effect; however the problem with this approach is that the metric properties may be lost, leading to loss of accuracy. Here, we study the underlying properties of such transforms and their effect on metric indexing mechanisms. We show there are some spaces where certain transforms may be applied without loss of accuracy, and further spaces where we can understand the engineering tradeoffs between accuracy and efficiency. We back these observations with experimental analysis. To highlight the value of the approach, we show three large spaces deriving from practical domains whose dimensionality prevents normal indexing techniques, but where the transforms applied give scalable access with a relatively small loss of accuracy.

Links

EF16_019/0000822, research and development project

Name: Centrum excelence pro kyberkriminalitu, kyberbezpečnost a ochranu kritických informačních infrastruktur

Citovat

CONNOR, Richard, Alan DEARLE, Vladimír MÍČ and Pavel ZEZULA. On the Application of Convex Transforms to Metric Search. Pattern Recognition Letters. 2020, vol. 138, October 2020, p. 563-570. ISSN 0167-8655. Available from: https://dx.doi.org/10.1016/j.patrec.2020.08.008.

@article{1673423,
   author = {Connor, Richard and Dearle, Alan and Míč, Vladimír and Zezula, Pavel},
   article_number = {October 2020},
   doi = {http://dx.doi.org/10.1016/j.patrec.2020.08.008},
   keywords = {similarity search; transformation of distance function; metric space; convex transform},
   language = {eng},
   issn = {0167-8655},
   journal = {Pattern Recognition Letters},
   title = {On the Application of Convex Transforms to Metric Search},
   url = {https://www.sciencedirect.com/science/article/abs/pii/S0167865520303068},
   volume = {138},
   year = {2020}
}

TY  - JOUR
ID  - 1673423
AU  - Connor, Richard - Dearle, Alan - Míč, Vladimír - Zezula, Pavel
PY  - 2020
TI  - On the Application of Convex Transforms to Metric Search
JF  - Pattern Recognition Letters
VL  - 138
IS  - October 2020
SP  - 563-570
EP  - 563-570
SN  - 01678655
KW  - similarity search
KW  - transformation of distance function
KW  - metric space
KW  - convex transform
UR  - https://www.sciencedirect.com/science/article/abs/pii/S0167865520303068
L2  - https://www.sciencedirect.com/science/article/abs/pii/S0167865520303068
N2  - Scalable similarity search in metric spaces relies on using the mathematical properties of the space in order to allow efficient querying. Most important in this context is the triangle inequality property, which can allow the majority of individual similarity comparisons to be avoided for a given query. However many important metric spaces, typically those with high dimensionality, are not amenable to such techniques. In the past convex transforms have been studied as a pragmatic mechanism which can overcome this effect; however the problem with this approach is that the metric properties may be lost, leading to loss of accuracy. Here, we study the underlying properties of such transforms and their effect on metric indexing mechanisms. We show there are some spaces where certain transforms may be applied without loss of accuracy, and further spaces where we can understand the engineering tradeoffs between accuracy and efficiency. We back these observations with experimental analysis. To highlight the value of the approach, we show three large spaces deriving from practical domains whose dimensionality prevents normal indexing techniques, but where the transforms applied give scalable access with a relatively small loss of accuracy.
ER  -

CONNOR, Richard, Alan DEARLE, Vladimír MÍČ and Pavel ZEZULA. On the Application of Convex Transforms to Metric Search. \textit{Pattern Recognition Letters}. 2020, vol.~138, October 2020, p.~563-570. ISSN~0167-8655. Available from: https://dx.doi.org/10.1016/j.patrec.2020.08.008.

Detailed Information on Publication Record