AÇAI: Ascent Similarity Caching with Approximate Indexes
T. Salem, G. Neglia, and D. Carra. 2021 33rd International Teletraffic Congress (ITC-33), page 1-9. Avignon, France, (August 2021)
Abstract
Similarity search is a key operation in multimedia retrieval systems and recommender systems, and it will play an important role also for future machine learning and augmented reality applications. When these systems need to serve large objects with tight delay constraints, edge servers close to the end-user can operate as similarity caches to speed up the retrieval. In this paper we present AÇAI, a new similarity caching policy which improves on the state of the art by using (i) an (approximate) index for the whole catalog to decide which objects to serve locally and which to retrieve from the remote server, and (ii) a mirror ascent algorithm to update the set of local objects with strong guarantees even when the request process does not exhibit any statistical regularity.
%0 Conference Paper
%1 sal21ITC33
%A Salem, Tareq Si
%A Neglia, Giovanni
%A Carra, Damiano
%B 2021 33rd International Teletraffic Congress (ITC-33)
%C Avignon, France
%D 2021
%K Augmented_reality Delays Machine_learning Mirrors Multimedia_systems Servers Task_analysis itc itc33
%P 1-9
%T AÇAI: Ascent Similarity Caching with Approximate Indexes
%U https://gitlab2.informatik.uni-wuerzburg.de/itc-conference/itc-conference-public/-/raw/master/itc33/sal21ITC33.pdf?inline=true
%X Similarity search is a key operation in multimedia retrieval systems and recommender systems, and it will play an important role also for future machine learning and augmented reality applications. When these systems need to serve large objects with tight delay constraints, edge servers close to the end-user can operate as similarity caches to speed up the retrieval. In this paper we present AÇAI, a new similarity caching policy which improves on the state of the art by using (i) an (approximate) index for the whole catalog to decide which objects to serve locally and which to retrieve from the remote server, and (ii) a mirror ascent algorithm to update the set of local objects with strong guarantees even when the request process does not exhibit any statistical regularity.
@inproceedings{sal21ITC33,
abstract = {Similarity search is a key operation in multimedia retrieval systems and recommender systems, and it will play an important role also for future machine learning and augmented reality applications. When these systems need to serve large objects with tight delay constraints, edge servers close to the end-user can operate as similarity caches to speed up the retrieval. In this paper we present AÇAI, a new similarity caching policy which improves on the state of the art by using (i) an (approximate) index for the whole catalog to decide which objects to serve locally and which to retrieve from the remote server, and (ii) a mirror ascent algorithm to update the set of local objects with strong guarantees even when the request process does not exhibit any statistical regularity.},
added-at = {2022-02-04T14:01:50.000+0100},
address = {Avignon, France},
author = {Salem, Tareq Si and Neglia, Giovanni and Carra, Damiano},
biburl = {https://www.bibsonomy.org/bibtex/2e9895977208b4de36de311f6c5bb57ea/itc},
booktitle = {2021 33rd International Teletraffic Congress (ITC-33)},
interhash = {e7e9fcb783e096e6aca566f80bab8197},
intrahash = {e9895977208b4de36de311f6c5bb57ea},
keywords = {Augmented_reality Delays Machine_learning Mirrors Multimedia_systems Servers Task_analysis itc itc33},
month = Aug,
pages = {1-9},
timestamp = {2022-02-04T14:01:50.000+0100},
title = {AÇAI: Ascent Similarity Caching with Approximate Indexes},
url = {https://gitlab2.informatik.uni-wuerzburg.de/itc-conference/itc-conference-public/-/raw/master/itc33/sal21ITC33.pdf?inline=true},
year = 2021
}