The Parameterized Complexity of Cascading Portfolio Scheduling

D 2019

The Parameterized Complexity of Cascading Portfolio Scheduling

EIBEN, Eduard, Robert GANIAN, Iyad KANJ and Stefan SZEIDER

Basic information

Original name

The Parameterized Complexity of Cascading Portfolio Scheduling

Authors

EIBEN, Eduard (703 Slovakia), Robert GANIAN (203 Czech Republic, guarantor, belonging to the institution), Iyad KANJ (840 United States of America) and Stefan SZEIDER (40 Austria)

Edition

USA, Advances in Neural Information Processing Systems 32 (NIPS 2019), p. 7666-7676, 11 pp. 2019

Publisher

Neural Information Processing Systems Foundation, Inc.

Other information

Language

English

Type of outcome

Stať ve sborníku

Field of Study

10201 Computer sciences, information science, bioinformatics

Country of publisher

United States of America

Confidentiality degree

není předmětem státního či obchodního tajemství

Publication form

electronic version available online

References:

URL

RIV identification code

RIV/00216224:14330/19:00113727

Organization unit

Faculty of Informatics

ISBN

978-1-7281-1044-8

UT WoS

000534424307066

Keywords in English

Parameterized Complexity

Abstract

V originále

Cascading portfolio scheduling is a static algorithm selection strategy which uses a sample of test instances to compute an optimal ordering (a cascading schedule) of a portfolio of available algorithms. The algorithms are then applied to each future instance according to this cascading schedule, until some algorithm in the schedule succeeds. Cascading algorithm scheduling has proven to be effective in several applications, including QBF solving and the generation of ImageNet classification models. It is known that the computation of an optimal cascading schedule in the offline phase is NP-hard. In this paper we study the parameterized complexity of this problem and establish its fixed-parameter tractability by utilizing structural properties of the success relation between algorithms and test instances. Our findings are significant as they reveal that in spite of the intractability of the problem in its general form, one can indeed exploit sparseness or density of the success relation to obtain non-trivial runtime guarantees for finding an optimal cascading schedule.

Citovat

EIBEN, Eduard, Robert GANIAN, Iyad KANJ and Stefan SZEIDER. The Parameterized Complexity of Cascading Portfolio Scheduling. Online. In Hanna M. Wallach and Hugo Larochelle and Alina Beygelzimer and Florence d'Alche-Buc and Emily B. Fox and Roman Garnett. Advances in Neural Information Processing Systems 32 (NIPS 2019). USA: Neural Information Processing Systems Foundation, Inc., 2019, p. 7666-7676. ISBN 978-1-7281-1044-8.

@inproceedings{1648298,
   author = {Eiben, Eduard and Ganian, Robert and Kanj, Iyad and Szeider, Stefan},
   address = {USA},
   booktitle = {Advances in Neural Information Processing Systems 32 (NIPS 2019)},
   editor = {Hanna M. Wallach and Hugo Larochelle and Alina Beygelzimer and Florence d'Alche-Buc and Emily B. Fox and Roman Garnett},
   keywords = {Parameterized Complexity},
   howpublished = {elektronická verze "online"},
   language = {eng},
   location = {USA},
   isbn = {978-1-7281-1044-8},
   pages = {7666-7676},
   publisher = {Neural Information Processing Systems Foundation, Inc.},
   title = {The Parameterized Complexity of Cascading Portfolio Scheduling},
   url = {http://papers.nips.cc/paper/8983-the-parameterized-complexity-of-cascading-portfolio-scheduling},
   year = {2019}
}

TY  - JOUR
ID  - 1648298
AU  - Eiben, Eduard - Ganian, Robert - Kanj, Iyad - Szeider, Stefan
PY  - 2019
TI  - The Parameterized Complexity of Cascading Portfolio Scheduling
PB  - Neural Information Processing Systems Foundation, Inc.
CY  - USA
SN  - 9781728110448
KW  - Parameterized Complexity
UR  - http://papers.nips.cc/paper/8983-the-parameterized-complexity-of-cascading-portfolio-scheduling
L2  - http://papers.nips.cc/paper/8983-the-parameterized-complexity-of-cascading-portfolio-scheduling
N2  - Cascading portfolio scheduling is a static algorithm selection strategy which uses a sample of test instances to compute an optimal ordering (a cascading schedule) of a portfolio of available algorithms. The algorithms are then applied to each future instance according to this cascading schedule, until some algorithm in the schedule succeeds. Cascading algorithm scheduling has proven to be effective in several applications, including QBF solving and the generation of ImageNet classification models. It is known that the computation of an optimal cascading schedule in the offline phase is NP-hard. In this paper we study the parameterized complexity of this problem and establish its fixed-parameter tractability by utilizing structural properties of the success relation between algorithms and test instances. Our findings are significant as they reveal that in spite of the intractability of the problem in its general form, one can indeed exploit sparseness or density of the success relation to obtain non-trivial runtime guarantees for finding an optimal cascading schedule.
ER  -

EIBEN, Eduard, Robert GANIAN, Iyad KANJ and Stefan SZEIDER. The Parameterized Complexity of Cascading Portfolio Scheduling. Online. In Hanna M. Wallach and Hugo Larochelle and Alina Beygelzimer and Florence d'Alche-Buc and Emily B. Fox and Roman Garnett. \textit{Advances in Neural Information Processing Systems 32 (NIPS 2019)}. USA: Neural Information Processing Systems Foundation, Inc., 2019, p.~7666-7676. ISBN~978-1-7281-1044-8.

Detailed Information on Publication Record