Submodular learning and covering with response-dependent costs

Sivan Sabato

doi:10.1007/978-3-319-46379-7_9

Submodular learning and covering with response-dependent costs

Sivan Sabato

Department of Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

We consider interactive learning and covering problems, in a setting where actions may incur different costs, depending on the response to the action. We propose a natural greedy algorithm for response-dependent costs. We bound the approximation factor of this greedy algorithm in active learning settings as well as in the general setting. We show that a different property of the cost function controls the approximation factor in each of these scenarios. We further show that in both settings, the approximation factor of this greedy algorithm is near-optimal among all greedy algorithms. Experiments demonstrate the advantages of the proposed algorithm in the response-dependent cost setting.

Original language	English
Title of host publication	Algorithmic Learning Theory - 27th International Conference, ALT 2016, Proceedings
Editors	Hans Ulrich Simon, Sandra Zilles, Ronald Ortner
Publisher	Springer Verlag
Pages	130-144
Number of pages	15
ISBN (Print)	9783319463780
DOIs	https://doi.org/10.1007/978-3-319-46379-7_9
State	Published - 1 Jan 2016
Event	27th International Conference on Algorithmic Learning Theory, ALT 2016 - Bari, Italy Duration: 19 Oct 2016 → 21 Oct 2016

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	9925 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	27th International Conference on Algorithmic Learning Theory, ALT 2016
Country/Territory	Italy
City	Bari
Period	19/10/16 → 21/10/16

Keywords

Interactive learning
Outcome costs
Submodular functions

ASJC Scopus subject areas

Theoretical Computer Science
General Computer Science

Access to Document

10.1007/978-3-319-46379-7_9

Cite this

Sabato, S. (2016). Submodular learning and covering with response-dependent costs. In H. U. Simon, S. Zilles, & R. Ortner (Eds.), Algorithmic Learning Theory - 27th International Conference, ALT 2016, Proceedings (pp. 130-144). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9925 LNAI). Springer Verlag. https://doi.org/10.1007/978-3-319-46379-7_9

Sabato, Sivan. / Submodular learning and covering with response-dependent costs. Algorithmic Learning Theory - 27th International Conference, ALT 2016, Proceedings. editor / Hans Ulrich Simon ; Sandra Zilles ; Ronald Ortner. Springer Verlag, 2016. pp. 130-144 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{3d4e9fc542084781899f90d357c074e3,

title = "Submodular learning and covering with response-dependent costs",

abstract = "We consider interactive learning and covering problems, in a setting where actions may incur different costs, depending on the response to the action. We propose a natural greedy algorithm for response-dependent costs. We bound the approximation factor of this greedy algorithm in active learning settings as well as in the general setting. We show that a different property of the cost function controls the approximation factor in each of these scenarios. We further show that in both settings, the approximation factor of this greedy algorithm is near-optimal among all greedy algorithms. Experiments demonstrate the advantages of the proposed algorithm in the response-dependent cost setting.",

keywords = "Interactive learning, Outcome costs, Submodular functions",

author = "Sivan Sabato",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing Switzerland 2016.; 27th International Conference on Algorithmic Learning Theory, ALT 2016 ; Conference date: 19-10-2016 Through 21-10-2016",

year = "2016",

month = jan,

day = "1",

doi = "10.1007/978-3-319-46379-7_9",

language = "English",

isbn = "9783319463780",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "130--144",

editor = "Simon, {Hans Ulrich} and Sandra Zilles and Ronald Ortner",

booktitle = "Algorithmic Learning Theory - 27th International Conference, ALT 2016, Proceedings",

address = "Germany",

}

Sabato, S 2016, Submodular learning and covering with response-dependent costs. in HU Simon, S Zilles & R Ortner (eds), Algorithmic Learning Theory - 27th International Conference, ALT 2016, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9925 LNAI, Springer Verlag, pp. 130-144, 27th International Conference on Algorithmic Learning Theory, ALT 2016, Bari, Italy, 19/10/16. https://doi.org/10.1007/978-3-319-46379-7_9

Submodular learning and covering with response-dependent costs. / Sabato, Sivan.
Algorithmic Learning Theory - 27th International Conference, ALT 2016, Proceedings. ed. / Hans Ulrich Simon; Sandra Zilles; Ronald Ortner. Springer Verlag, 2016. p. 130-144 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9925 LNAI).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Submodular learning and covering with response-dependent costs

AU - Sabato, Sivan

N1 - Publisher Copyright: © Springer International Publishing Switzerland 2016.

PY - 2016/1/1

Y1 - 2016/1/1

N2 - We consider interactive learning and covering problems, in a setting where actions may incur different costs, depending on the response to the action. We propose a natural greedy algorithm for response-dependent costs. We bound the approximation factor of this greedy algorithm in active learning settings as well as in the general setting. We show that a different property of the cost function controls the approximation factor in each of these scenarios. We further show that in both settings, the approximation factor of this greedy algorithm is near-optimal among all greedy algorithms. Experiments demonstrate the advantages of the proposed algorithm in the response-dependent cost setting.

AB - We consider interactive learning and covering problems, in a setting where actions may incur different costs, depending on the response to the action. We propose a natural greedy algorithm for response-dependent costs. We bound the approximation factor of this greedy algorithm in active learning settings as well as in the general setting. We show that a different property of the cost function controls the approximation factor in each of these scenarios. We further show that in both settings, the approximation factor of this greedy algorithm is near-optimal among all greedy algorithms. Experiments demonstrate the advantages of the proposed algorithm in the response-dependent cost setting.

KW - Interactive learning

KW - Outcome costs

KW - Submodular functions

UR - http://www.scopus.com/inward/record.url?scp=84994111990&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-46379-7_9

DO - 10.1007/978-3-319-46379-7_9

M3 - Conference contribution

AN - SCOPUS:84994111990

SN - 9783319463780

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 130

EP - 144

BT - Algorithmic Learning Theory - 27th International Conference, ALT 2016, Proceedings

A2 - Simon, Hans Ulrich

A2 - Zilles, Sandra

A2 - Ortner, Ronald

PB - Springer Verlag

T2 - 27th International Conference on Algorithmic Learning Theory, ALT 2016

Y2 - 19 October 2016 through 21 October 2016

ER -

Sabato S. Submodular learning and covering with response-dependent costs. In Simon HU, Zilles S, Ortner R, editors, Algorithmic Learning Theory - 27th International Conference, ALT 2016, Proceedings. Springer Verlag. 2016. p. 130-144. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-46379-7_9

Submodular learning and covering with response-dependent costs

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this