Large-scale training of SVMs with automata kernels

Cyril Allauzen, Corinna Cortes, Mehryar Mohri

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

Abstract

This paper presents a novel application of automata algorithms to machine learning. It introduces the first optimization solution for support vector machines used with sequence kernels that is purely based on weighted automata and transducer algorithms, without requiring any specific solver. The algorithms presented apply to a family of kernels covering all those commonly used in text and speech processing or computational biology. We show that these algorithms have significantly better computational complexity than previous ones and report the results of large-scale experiments demonstrating a dramatic reduction of the training time, typically by several orders of magnitude.

Original languageEnglish (US)
Title of host publicationImplementation and Application of Automata - 15th International Conference, CIAA 2010, Revised Selected Papers
Pages17-27
Number of pages11
Volume6482 LNCS
DOIs
StatePublished - 2011
Event15th International Conference on Implementation and Application of Automata, CIAA 2010 - Winnipeg, MB, Canada
Duration: Aug 12 2010Aug 15 2010

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume6482 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other15th International Conference on Implementation and Application of Automata, CIAA 2010
CountryCanada
CityWinnipeg, MB
Period8/12/108/15/10

Fingerprint

Automata
kernel
Text processing
Weighted Automata
Text Processing
Speech Processing
Speech processing
Computational Biology
Transducer
Support vector machines
Learning systems
Transducers
Computational complexity
Support Vector Machine
Machine Learning
Computational Complexity
Covering
Training
Optimization
Experiment

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Allauzen, C., Cortes, C., & Mohri, M. (2011). Large-scale training of SVMs with automata kernels. In Implementation and Application of Automata - 15th International Conference, CIAA 2010, Revised Selected Papers (Vol. 6482 LNCS, pp. 17-27). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6482 LNCS). https://doi.org/10.1007/978-3-642-18098-9_3

Large-scale training of SVMs with automata kernels. / Allauzen, Cyril; Cortes, Corinna; Mohri, Mehryar.

Implementation and Application of Automata - 15th International Conference, CIAA 2010, Revised Selected Papers. Vol. 6482 LNCS 2011. p. 17-27 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6482 LNCS).

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

Allauzen, C, Cortes, C & Mohri, M 2011, Large-scale training of SVMs with automata kernels. in Implementation and Application of Automata - 15th International Conference, CIAA 2010, Revised Selected Papers. vol. 6482 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 6482 LNCS, pp. 17-27, 15th International Conference on Implementation and Application of Automata, CIAA 2010, Winnipeg, MB, Canada, 8/12/10. https://doi.org/10.1007/978-3-642-18098-9_3
Allauzen C, Cortes C, Mohri M. Large-scale training of SVMs with automata kernels. In Implementation and Application of Automata - 15th International Conference, CIAA 2010, Revised Selected Papers. Vol. 6482 LNCS. 2011. p. 17-27. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-18098-9_3
Allauzen, Cyril ; Cortes, Corinna ; Mohri, Mehryar. / Large-scale training of SVMs with automata kernels. Implementation and Application of Automata - 15th International Conference, CIAA 2010, Revised Selected Papers. Vol. 6482 LNCS 2011. pp. 17-27 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{a8aa9e41ae304abb90e409fd6db6a1de,
title = "Large-scale training of SVMs with automata kernels",
abstract = "This paper presents a novel application of automata algorithms to machine learning. It introduces the first optimization solution for support vector machines used with sequence kernels that is purely based on weighted automata and transducer algorithms, without requiring any specific solver. The algorithms presented apply to a family of kernels covering all those commonly used in text and speech processing or computational biology. We show that these algorithms have significantly better computational complexity than previous ones and report the results of large-scale experiments demonstrating a dramatic reduction of the training time, typically by several orders of magnitude.",
author = "Cyril Allauzen and Corinna Cortes and Mehryar Mohri",
year = "2011",
doi = "10.1007/978-3-642-18098-9_3",
language = "English (US)",
isbn = "9783642180972",
volume = "6482 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "17--27",
booktitle = "Implementation and Application of Automata - 15th International Conference, CIAA 2010, Revised Selected Papers",

}

TY - GEN

T1 - Large-scale training of SVMs with automata kernels

AU - Allauzen, Cyril

AU - Cortes, Corinna

AU - Mohri, Mehryar

PY - 2011

Y1 - 2011

N2 - This paper presents a novel application of automata algorithms to machine learning. It introduces the first optimization solution for support vector machines used with sequence kernels that is purely based on weighted automata and transducer algorithms, without requiring any specific solver. The algorithms presented apply to a family of kernels covering all those commonly used in text and speech processing or computational biology. We show that these algorithms have significantly better computational complexity than previous ones and report the results of large-scale experiments demonstrating a dramatic reduction of the training time, typically by several orders of magnitude.

AB - This paper presents a novel application of automata algorithms to machine learning. It introduces the first optimization solution for support vector machines used with sequence kernels that is purely based on weighted automata and transducer algorithms, without requiring any specific solver. The algorithms presented apply to a family of kernels covering all those commonly used in text and speech processing or computational biology. We show that these algorithms have significantly better computational complexity than previous ones and report the results of large-scale experiments demonstrating a dramatic reduction of the training time, typically by several orders of magnitude.

UR - http://www.scopus.com/inward/record.url?scp=79951594305&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79951594305&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-18098-9_3

DO - 10.1007/978-3-642-18098-9_3

M3 - Conference contribution

SN - 9783642180972

VL - 6482 LNCS

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

SP - 17

EP - 27

BT - Implementation and Application of Automata - 15th International Conference, CIAA 2010, Revised Selected Papers

ER -