OU Portal
Log In
Welcome
Applicants
Z6_60GI02O0O8IDC0QEJUJ26TJDI4
Error:
Javascript is disabled in this browser. This page requires Javascript. Modify your browser's settings to allow Javascript to execute. See your browser's documentation for specific instructions.
{}
Close
Publikační činnost
Probíhá načítání, čekejte prosím...
publicationId :
tempRecordId :
actionDispatchIndex :
navigationBranch :
pageMode :
tabSelected :
isRivValid :
Record type:
stať ve sborníku (D)
Home Department:
Ústav pro výzkum a aplikace fuzzy modelování (94410)
Title:
Graph Neural Networks for Scheduling of SMT Solvers
Citace
HŮLA, J., MOJŽÍŠEK, D. a Janota, M. Graph Neural Networks for Scheduling of SMT Solvers.
In:
2021 IEEE 33RD INTERNATIONAL CONFERENCE ON TOOLS WITH ARTIFICIAL INTELLIGENCE (ICTAI 2021) 2021-11-01 Online.
Los Alamitos, USA: IEEE, 2021. s. 447-451. ISBN 978-1-6654-0898-1.
Subtitle
Publication year:
2021
Obor:
Obecná matematika
Number of pages:
5
Page from:
447
Page to:
451
Form of publication:
Elektronická verze
ISBN code:
978-1-6654-0898-1
ISSN code:
Proceedings title:
2021 IEEE 33RD INTERNATIONAL CONFERENCE ON TOOLS WITH ARTIFICIAL INTELLIGENCE (ICTAI 2021)
Proceedings:
Mezinárodní
Publisher name:
IEEE
Place of publishing:
Los Alamitos, USA
Country of Publication:
Sborník vydaný v zahraničí
Název konference:
Místo konání konference:
Online
Datum zahájení konference:
Typ akce podle státní
příslušnosti účastníků:
Celosvětová akce
WoS code:
000747482300064
EID:
2-s2.0-85123953302
Key words in English:
SMT, graph neural networks, scheduling
Annotation in original language:
This paper develops an approach to the scheduling of solvers in the domain of Satisfiability Modulo Theories (SMT) using a Graph Neural Network (GNN). In contrast to related methods, GNNs do not require manual feature design as they enable discovering relevant features in the raw data. We train them to predict the effectivity of individual solvers on a given problem. Rather than choosing only one solver with the best prediction, we schedule the solvers by ordering them according to the predicted runtime and dividing the overall runtime into all solvers uniformly. We compare our approach to several baselines. In the selected benchmarks, we show a substantial improvement over these baselines in terms of the number of solved problems and overall solving time.
Annotation in english language:
References
Reference
R01:
RIV/61988987:17610/21:A2402MER
Complementary Content
Deferred Modules
${title}
${badge}
${loading}
Deferred Modules