Learning DFA from simple examples

Rajesh Parekh, Vasant Honavar

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

16 Citations (Scopus)

Abstract

We present a framework for learning DFA from simple examples. We show that efficient PAC learning of DFA is possible if the class of distributions is restricted to simple distributions where a teacher might choose examples based on the knowledge of the target concept. This answers an interesting variant of an open research question posed in Pitt's seminal paper: Are DFA's PAC-identifiable if examples are drawn from the uniform distribution, or some other known simple distribution? Our approach uses the RPNI algorithm for learning DFA from labeled exampies. In particular, we describe an efficient learning algorithm for exact learning of the target DFA with high probability when a bound on the number of states (N) of the target DFA is known in advance. When N is not known, we show how this algorithm can be used for efficient PAC learning of DFAs.

Original languageEnglish (US)
Title of host publicationAlgorithmic Learning Theory - 8th International Workshop, ALT 1997, Proceedings
EditorsMing Li, Akira Maruoka
PublisherSpringer Verlag
Pages116-131
Number of pages16
ISBN (Print)3540635777, 9783540635772
DOIs
StatePublished - Jan 1 1997
Event8th International Workshop on Algorithmic Learning Theory, ALT 1997 - Sendai, Japan
Duration: Oct 6 1997Oct 8 1997

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume1316
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other8th International Workshop on Algorithmic Learning Theory, ALT 1997
CountryJapan
CitySendai
Period10/6/9710/8/97

Fingerprint

PAC Learning
Target
Learning algorithms
Uniform distribution
Learning Algorithm
Efficient Algorithms
Choose
Learning

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Parekh, R., & Honavar, V. (1997). Learning DFA from simple examples. In M. Li, & A. Maruoka (Eds.), Algorithmic Learning Theory - 8th International Workshop, ALT 1997, Proceedings (pp. 116-131). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1316). Springer Verlag. https://doi.org/10.1007/3-540-63577-7_39
Parekh, Rajesh ; Honavar, Vasant. / Learning DFA from simple examples. Algorithmic Learning Theory - 8th International Workshop, ALT 1997, Proceedings. editor / Ming Li ; Akira Maruoka. Springer Verlag, 1997. pp. 116-131 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{41fcf4aca1a24a36b7d1b8fae94baff6,
title = "Learning DFA from simple examples",
abstract = "We present a framework for learning DFA from simple examples. We show that efficient PAC learning of DFA is possible if the class of distributions is restricted to simple distributions where a teacher might choose examples based on the knowledge of the target concept. This answers an interesting variant of an open research question posed in Pitt's seminal paper: Are DFA's PAC-identifiable if examples are drawn from the uniform distribution, or some other known simple distribution? Our approach uses the RPNI algorithm for learning DFA from labeled exampies. In particular, we describe an efficient learning algorithm for exact learning of the target DFA with high probability when a bound on the number of states (N) of the target DFA is known in advance. When N is not known, we show how this algorithm can be used for efficient PAC learning of DFAs.",
author = "Rajesh Parekh and Vasant Honavar",
year = "1997",
month = "1",
day = "1",
doi = "10.1007/3-540-63577-7_39",
language = "English (US)",
isbn = "3540635777",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "116--131",
editor = "Ming Li and Akira Maruoka",
booktitle = "Algorithmic Learning Theory - 8th International Workshop, ALT 1997, Proceedings",
address = "Germany",

}

Parekh, R & Honavar, V 1997, Learning DFA from simple examples. in M Li & A Maruoka (eds), Algorithmic Learning Theory - 8th International Workshop, ALT 1997, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 1316, Springer Verlag, pp. 116-131, 8th International Workshop on Algorithmic Learning Theory, ALT 1997, Sendai, Japan, 10/6/97. https://doi.org/10.1007/3-540-63577-7_39

Learning DFA from simple examples. / Parekh, Rajesh; Honavar, Vasant.

Algorithmic Learning Theory - 8th International Workshop, ALT 1997, Proceedings. ed. / Ming Li; Akira Maruoka. Springer Verlag, 1997. p. 116-131 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1316).

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

TY - GEN

T1 - Learning DFA from simple examples

AU - Parekh, Rajesh

AU - Honavar, Vasant

PY - 1997/1/1

Y1 - 1997/1/1

N2 - We present a framework for learning DFA from simple examples. We show that efficient PAC learning of DFA is possible if the class of distributions is restricted to simple distributions where a teacher might choose examples based on the knowledge of the target concept. This answers an interesting variant of an open research question posed in Pitt's seminal paper: Are DFA's PAC-identifiable if examples are drawn from the uniform distribution, or some other known simple distribution? Our approach uses the RPNI algorithm for learning DFA from labeled exampies. In particular, we describe an efficient learning algorithm for exact learning of the target DFA with high probability when a bound on the number of states (N) of the target DFA is known in advance. When N is not known, we show how this algorithm can be used for efficient PAC learning of DFAs.

AB - We present a framework for learning DFA from simple examples. We show that efficient PAC learning of DFA is possible if the class of distributions is restricted to simple distributions where a teacher might choose examples based on the knowledge of the target concept. This answers an interesting variant of an open research question posed in Pitt's seminal paper: Are DFA's PAC-identifiable if examples are drawn from the uniform distribution, or some other known simple distribution? Our approach uses the RPNI algorithm for learning DFA from labeled exampies. In particular, we describe an efficient learning algorithm for exact learning of the target DFA with high probability when a bound on the number of states (N) of the target DFA is known in advance. When N is not known, we show how this algorithm can be used for efficient PAC learning of DFAs.

UR - http://www.scopus.com/inward/record.url?scp=84958043718&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84958043718&partnerID=8YFLogxK

U2 - 10.1007/3-540-63577-7_39

DO - 10.1007/3-540-63577-7_39

M3 - Conference contribution

AN - SCOPUS:84958043718

SN - 3540635777

SN - 9783540635772

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

SP - 116

EP - 131

BT - Algorithmic Learning Theory - 8th International Workshop, ALT 1997, Proceedings

A2 - Li, Ming

A2 - Maruoka, Akira

PB - Springer Verlag

ER -

Parekh R, Honavar V. Learning DFA from simple examples. In Li M, Maruoka A, editors, Algorithmic Learning Theory - 8th International Workshop, ALT 1997, Proceedings. Springer Verlag. 1997. p. 116-131. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/3-540-63577-7_39