Learning by extended statistical queries and its relation to PAC learning

Eli Shamir; Clara Shwartzman

doi:10.1007/3-540-59119-2_191

Learning by extended statistical queries and its relation to PAC learning

Eli Shamir, Clara Shwartzman

Einstein Institute of Mathematics

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

5 Scopus citations

Abstract

PAC learning from examples is factored so that (i) the membership queries axe used to evaluate empirically "statistical queries" -certain expectations of ftmctionals involving the unknown target, (ii) approximate value of these statistical queries are used to compute an output - an approximation of the target. K earns' original formulation of statistical queries [we use the abbreviation SQ], is extended here to include as SQ functionals of arbitrary range and order higher than one - second order being the most useful addition. This enables us to capture more ground for casting efficient PAC learning algorithms in SQ form: The important Kushilevitz-Mansour Fourier - based algorithm has an SQ rendition, as well as its derivatives, e.g. Jackson's recent DNF learning. Efficient evaluation of extended SQ by membership queries, if possible at all, becomes quite intricate. We show, however, that it is usually robust under classification noise.

Original language	English
Title of host publication	Computational Learning Theory - 2nd European Conference, EuroCOLT 1995, Proceedings
Editors	Paul Vitanyi
Publisher	Springer Verlag
Pages	357-366
Number of pages	10
ISBN (Print)	9783540591191
DOIs	https://doi.org/10.1007/3-540-59119-2_191
State	Published - 1995
Event	2nd European Conference on Computational Learning Theory, EuroCOLT 1995 - Barcelona, Spain Duration: 13 Mar 1995 → 15 Mar 1995

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	904
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	2nd European Conference on Computational Learning Theory, EuroCOLT 1995
Country/Territory	Spain
City	Barcelona
Period	13/03/95 → 15/03/95

Bibliographical note

Publisher Copyright:
© Springer-Verlag Berlin Heidelberg 1995.

Access to Document

10.1007/3-540-59119-2_191

Cite this

Shamir, E., & Shwartzman, C. (1995). Learning by extended statistical queries and its relation to PAC learning. In P. Vitanyi (Ed.), Computational Learning Theory - 2nd European Conference, EuroCOLT 1995, Proceedings (pp. 357-366). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 904). Springer Verlag. https://doi.org/10.1007/3-540-59119-2_191

Shamir, Eli ; Shwartzman, Clara. / Learning by extended statistical queries and its relation to PAC learning. Computational Learning Theory - 2nd European Conference, EuroCOLT 1995, Proceedings. editor / Paul Vitanyi. Springer Verlag, 1995. pp. 357-366 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{0d8afe62c0e74740b391314400392839,

title = "Learning by extended statistical queries and its relation to PAC learning",

abstract = "PAC learning from examples is factored so that (i) the membership queries axe used to evaluate empirically {"}statistical queries{"} -certain expectations of ftmctionals involving the unknown target, (ii) approximate value of these statistical queries are used to compute an output - an approximation of the target. K earns' original formulation of statistical queries [we use the abbreviation SQ], is extended here to include as SQ functionals of arbitrary range and order higher than one - second order being the most useful addition. This enables us to capture more ground for casting efficient PAC learning algorithms in SQ form: The important Kushilevitz-Mansour Fourier - based algorithm has an SQ rendition, as well as its derivatives, e.g. Jackson's recent DNF learning. Efficient evaluation of extended SQ by membership queries, if possible at all, becomes quite intricate. We show, however, that it is usually robust under classification noise.",

author = "Eli Shamir and Clara Shwartzman",

note = "Publisher Copyright: {\textcopyright} Springer-Verlag Berlin Heidelberg 1995.; 2nd European Conference on Computational Learning Theory, EuroCOLT 1995 ; Conference date: 13-03-1995 Through 15-03-1995",

year = "1995",

doi = "10.1007/3-540-59119-2_191",

language = "אנגלית",

isbn = "9783540591191",

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

publisher = "Springer Verlag",

pages = "357--366",

editor = "Paul Vitanyi",

booktitle = "Computational Learning Theory - 2nd European Conference, EuroCOLT 1995, Proceedings",

address = "גרמניה",

}

Shamir, E & Shwartzman, C 1995, Learning by extended statistical queries and its relation to PAC learning. in P Vitanyi (ed.), Computational Learning Theory - 2nd European Conference, EuroCOLT 1995, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 904, Springer Verlag, pp. 357-366, 2nd European Conference on Computational Learning Theory, EuroCOLT 1995, Barcelona, Spain, 13/03/95. https://doi.org/10.1007/3-540-59119-2_191

Learning by extended statistical queries and its relation to PAC learning. / Shamir, Eli; Shwartzman, Clara.
Computational Learning Theory - 2nd European Conference, EuroCOLT 1995, Proceedings. ed. / Paul Vitanyi. Springer Verlag, 1995. p. 357-366 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 904).

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

TY - GEN

T1 - Learning by extended statistical queries and its relation to PAC learning

AU - Shamir, Eli

AU - Shwartzman, Clara

PY - 1995

Y1 - 1995

N2 - PAC learning from examples is factored so that (i) the membership queries axe used to evaluate empirically "statistical queries" -certain expectations of ftmctionals involving the unknown target, (ii) approximate value of these statistical queries are used to compute an output - an approximation of the target. K earns' original formulation of statistical queries [we use the abbreviation SQ], is extended here to include as SQ functionals of arbitrary range and order higher than one - second order being the most useful addition. This enables us to capture more ground for casting efficient PAC learning algorithms in SQ form: The important Kushilevitz-Mansour Fourier - based algorithm has an SQ rendition, as well as its derivatives, e.g. Jackson's recent DNF learning. Efficient evaluation of extended SQ by membership queries, if possible at all, becomes quite intricate. We show, however, that it is usually robust under classification noise.

AB - PAC learning from examples is factored so that (i) the membership queries axe used to evaluate empirically "statistical queries" -certain expectations of ftmctionals involving the unknown target, (ii) approximate value of these statistical queries are used to compute an output - an approximation of the target. K earns' original formulation of statistical queries [we use the abbreviation SQ], is extended here to include as SQ functionals of arbitrary range and order higher than one - second order being the most useful addition. This enables us to capture more ground for casting efficient PAC learning algorithms in SQ form: The important Kushilevitz-Mansour Fourier - based algorithm has an SQ rendition, as well as its derivatives, e.g. Jackson's recent DNF learning. Efficient evaluation of extended SQ by membership queries, if possible at all, becomes quite intricate. We show, however, that it is usually robust under classification noise.

UR - http://www.scopus.com/inward/record.url?scp=21844490235&partnerID=8YFLogxK

U2 - 10.1007/3-540-59119-2_191

DO - 10.1007/3-540-59119-2_191

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

AN - SCOPUS:21844490235

SN - 9783540591191

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

SP - 357

EP - 366

BT - Computational Learning Theory - 2nd European Conference, EuroCOLT 1995, Proceedings

A2 - Vitanyi, Paul

PB - Springer Verlag

T2 - 2nd European Conference on Computational Learning Theory, EuroCOLT 1995

Y2 - 13 March 1995 through 15 March 1995

ER -

Shamir E, Shwartzman C. Learning by extended statistical queries and its relation to PAC learning. In Vitanyi P, editor, Computational Learning Theory - 2nd European Conference, EuroCOLT 1995, Proceedings. Springer Verlag. 1995. p. 357-366. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/3-540-59119-2_191

Learning by extended statistical queries and its relation to PAC learning

Abstract

Publication series

Conference

Bibliographical note

Access to Document

Other files and links

Fingerprint

Cite this