Efficient algorithms for combinatorial online prediction

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

We study online linear optimization problems over concept classes which are defined in some combinatorial ways. Typically, those concept classes contain finite but exponentially many concepts and hence the complexity issue arises. In this paper, we survey some recent results on universal and efficient implementations of low-regret algorithmic frameworks such as Follow the Regularized Leader (FTRL) and Follow the Perturbed Leader (FPL).

Original languageEnglish
Title of host publicationAlgorithmic Learning Theory - 24th International Conference, ALT 2013, Proceedings
Pages22-32
Number of pages11
DOIs
Publication statusPublished - Nov 18 2013
Event24th International Conference on Algorithmic Learning Theory, ALT 2013 - Singapore, Singapore
Duration: Oct 6 2013Oct 9 2013

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8139 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other24th International Conference on Algorithmic Learning Theory, ALT 2013
CountrySingapore
CitySingapore
Period10/6/1310/9/13

Fingerprint

Efficient Algorithms
Prediction
Online Optimization
Linear Optimization
Regret
Efficient Implementation
Optimization Problem
Concepts
Class
Framework

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Takimoto, E., & Hatano, K. (2013). Efficient algorithms for combinatorial online prediction. In Algorithmic Learning Theory - 24th International Conference, ALT 2013, Proceedings (pp. 22-32). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8139 LNAI). https://doi.org/10.1007/978-3-642-40935-6_3

Efficient algorithms for combinatorial online prediction. / Takimoto, Eiji; Hatano, Kohei.

Algorithmic Learning Theory - 24th International Conference, ALT 2013, Proceedings. 2013. p. 22-32 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8139 LNAI).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Takimoto, E & Hatano, K 2013, Efficient algorithms for combinatorial online prediction. in Algorithmic Learning Theory - 24th International Conference, ALT 2013, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 8139 LNAI, pp. 22-32, 24th International Conference on Algorithmic Learning Theory, ALT 2013, Singapore, Singapore, 10/6/13. https://doi.org/10.1007/978-3-642-40935-6_3
Takimoto E, Hatano K. Efficient algorithms for combinatorial online prediction. In Algorithmic Learning Theory - 24th International Conference, ALT 2013, Proceedings. 2013. p. 22-32. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-40935-6_3
Takimoto, Eiji ; Hatano, Kohei. / Efficient algorithms for combinatorial online prediction. Algorithmic Learning Theory - 24th International Conference, ALT 2013, Proceedings. 2013. pp. 22-32 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{2ba64900e73b4d8596bc8aca70529be7,
title = "Efficient algorithms for combinatorial online prediction",
abstract = "We study online linear optimization problems over concept classes which are defined in some combinatorial ways. Typically, those concept classes contain finite but exponentially many concepts and hence the complexity issue arises. In this paper, we survey some recent results on universal and efficient implementations of low-regret algorithmic frameworks such as Follow the Regularized Leader (FTRL) and Follow the Perturbed Leader (FPL).",
author = "Eiji Takimoto and Kohei Hatano",
year = "2013",
month = "11",
day = "18",
doi = "10.1007/978-3-642-40935-6_3",
language = "English",
isbn = "9783642409349",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "22--32",
booktitle = "Algorithmic Learning Theory - 24th International Conference, ALT 2013, Proceedings",

}

TY - GEN

T1 - Efficient algorithms for combinatorial online prediction

AU - Takimoto, Eiji

AU - Hatano, Kohei

PY - 2013/11/18

Y1 - 2013/11/18

N2 - We study online linear optimization problems over concept classes which are defined in some combinatorial ways. Typically, those concept classes contain finite but exponentially many concepts and hence the complexity issue arises. In this paper, we survey some recent results on universal and efficient implementations of low-regret algorithmic frameworks such as Follow the Regularized Leader (FTRL) and Follow the Perturbed Leader (FPL).

AB - We study online linear optimization problems over concept classes which are defined in some combinatorial ways. Typically, those concept classes contain finite but exponentially many concepts and hence the complexity issue arises. In this paper, we survey some recent results on universal and efficient implementations of low-regret algorithmic frameworks such as Follow the Regularized Leader (FTRL) and Follow the Perturbed Leader (FPL).

UR - http://www.scopus.com/inward/record.url?scp=84887432263&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84887432263&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-40935-6_3

DO - 10.1007/978-3-642-40935-6_3

M3 - Conference contribution

AN - SCOPUS:84887432263

SN - 9783642409349

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

SP - 22

EP - 32

BT - Algorithmic Learning Theory - 24th International Conference, ALT 2013, Proceedings

ER -