On the relationship between models for learning in helpful environments

Rajesh Parekh, Vasant Honavar

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

5 Citations (Scopus)

Abstract

The PAC and other equivalent learning models are widely accepted models for polynomial learnability of concept classes. However, negative results abound in the PAC learning framework (concept classes such as deterministic finite state automata (DFA) are not efficiently learnable in the PAC model). The PAC model’s requirement of learnability under all conceivable distributions could be considered too stringent a restriction for practical applications. Several models for learning in more helpful environments have been proposed in the literature including: learning from example based queries [2], online learning allowing a bounded number of mistakes [14], learning with the help of teaching sets [7], learning from characteristic sets [5], and learning from simple examples [12,4]. Several concept classes that are not learnable in the standard PAC model have been shown to be learnable in these models In this paper we identify the relationships between these different learning models. We also address the issue of unnatural collusion between the teacher and the learner that can potentially trivialize the task of learning in helpful environments.

Original languageEnglish (US)
Title of host publicationGrammatical Inference
Subtitle of host publicationAlgorithms and Applications - 5th International Colloquium, ICGI 2000, Proceedings
EditorsArlindo L. Oliveira
PublisherSpringer Verlag
Pages207-220
Number of pages14
ISBN (Print)9783540452577
StatePublished - Jan 1 2000
Event5th International Colloquium on Grammatical Inference, ICGI 2000 - Lisbon, Portugal
Duration: Sep 11 2000Sep 13 2000

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume1891
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other5th International Colloquium on Grammatical Inference, ICGI 2000
CountryPortugal
CityLisbon
Period9/11/009/13/00

Fingerprint

Learnability
Model
PAC Learning
Characteristic Set
Collusion
Finite State Automata
Online Learning
Relationships
Learning
Finite automata
Teaching
Polynomials
Query
Restriction
Polynomial
Requirements
Concepts
Class
Framework
Standards

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Parekh, R., & Honavar, V. (2000). On the relationship between models for learning in helpful environments. In A. L. Oliveira (Ed.), Grammatical Inference: Algorithms and Applications - 5th International Colloquium, ICGI 2000, Proceedings (pp. 207-220). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1891). Springer Verlag.
Parekh, Rajesh ; Honavar, Vasant. / On the relationship between models for learning in helpful environments. Grammatical Inference: Algorithms and Applications - 5th International Colloquium, ICGI 2000, Proceedings. editor / Arlindo L. Oliveira. Springer Verlag, 2000. pp. 207-220 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{fd96029b549847a39a46c82a233e93b6,
title = "On the relationship between models for learning in helpful environments",
abstract = "The PAC and other equivalent learning models are widely accepted models for polynomial learnability of concept classes. However, negative results abound in the PAC learning framework (concept classes such as deterministic finite state automata (DFA) are not efficiently learnable in the PAC model). The PAC model’s requirement of learnability under all conceivable distributions could be considered too stringent a restriction for practical applications. Several models for learning in more helpful environments have been proposed in the literature including: learning from example based queries [2], online learning allowing a bounded number of mistakes [14], learning with the help of teaching sets [7], learning from characteristic sets [5], and learning from simple examples [12,4]. Several concept classes that are not learnable in the standard PAC model have been shown to be learnable in these models In this paper we identify the relationships between these different learning models. We also address the issue of unnatural collusion between the teacher and the learner that can potentially trivialize the task of learning in helpful environments.",
author = "Rajesh Parekh and Vasant Honavar",
year = "2000",
month = "1",
day = "1",
language = "English (US)",
isbn = "9783540452577",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "207--220",
editor = "Oliveira, {Arlindo L.}",
booktitle = "Grammatical Inference",
address = "Germany",

}

Parekh, R & Honavar, V 2000, On the relationship between models for learning in helpful environments. in AL Oliveira (ed.), Grammatical Inference: Algorithms and Applications - 5th International Colloquium, ICGI 2000, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 1891, Springer Verlag, pp. 207-220, 5th International Colloquium on Grammatical Inference, ICGI 2000, Lisbon, Portugal, 9/11/00.

On the relationship between models for learning in helpful environments. / Parekh, Rajesh; Honavar, Vasant.

Grammatical Inference: Algorithms and Applications - 5th International Colloquium, ICGI 2000, Proceedings. ed. / Arlindo L. Oliveira. Springer Verlag, 2000. p. 207-220 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1891).

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

TY - GEN

T1 - On the relationship between models for learning in helpful environments

AU - Parekh, Rajesh

AU - Honavar, Vasant

PY - 2000/1/1

Y1 - 2000/1/1

N2 - The PAC and other equivalent learning models are widely accepted models for polynomial learnability of concept classes. However, negative results abound in the PAC learning framework (concept classes such as deterministic finite state automata (DFA) are not efficiently learnable in the PAC model). The PAC model’s requirement of learnability under all conceivable distributions could be considered too stringent a restriction for practical applications. Several models for learning in more helpful environments have been proposed in the literature including: learning from example based queries [2], online learning allowing a bounded number of mistakes [14], learning with the help of teaching sets [7], learning from characteristic sets [5], and learning from simple examples [12,4]. Several concept classes that are not learnable in the standard PAC model have been shown to be learnable in these models In this paper we identify the relationships between these different learning models. We also address the issue of unnatural collusion between the teacher and the learner that can potentially trivialize the task of learning in helpful environments.

AB - The PAC and other equivalent learning models are widely accepted models for polynomial learnability of concept classes. However, negative results abound in the PAC learning framework (concept classes such as deterministic finite state automata (DFA) are not efficiently learnable in the PAC model). The PAC model’s requirement of learnability under all conceivable distributions could be considered too stringent a restriction for practical applications. Several models for learning in more helpful environments have been proposed in the literature including: learning from example based queries [2], online learning allowing a bounded number of mistakes [14], learning with the help of teaching sets [7], learning from characteristic sets [5], and learning from simple examples [12,4]. Several concept classes that are not learnable in the standard PAC model have been shown to be learnable in these models In this paper we identify the relationships between these different learning models. We also address the issue of unnatural collusion between the teacher and the learner that can potentially trivialize the task of learning in helpful environments.

UR - http://www.scopus.com/inward/record.url?scp=84974715465&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84974715465&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84974715465

SN - 9783540452577

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

SP - 207

EP - 220

BT - Grammatical Inference

A2 - Oliveira, Arlindo L.

PB - Springer Verlag

ER -

Parekh R, Honavar V. On the relationship between models for learning in helpful environments. In Oliveira AL, editor, Grammatical Inference: Algorithms and Applications - 5th International Colloquium, ICGI 2000, Proceedings. Springer Verlag. 2000. p. 207-220. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).