On hardness of learning intersection of two halfspaces

Subhash Khot, Rishi Saket

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

Abstract

We show that unless NP = RP, it is hard to (even) weakly PAC-learn intersection of two halfspaces in Rn using a hypothesis which is a, function of up to ℓ linear threshold functions for any integer ℓ. Specifically, we show that for every integer ℓ and an arbitrarily small constant ε > 0, unless NP = RP, no polynomial time algorithm can distinguish whether there is an intersection of two halfspaces that correctly classifies a given set of labeled points in Rn, or whether any function of ℓ linear threshold functions can correctly classify at most 1/2+ ε fraction of the points.

Original languageEnglish (US)
Title of host publicationSTOC'08: Proceedings of the 2008 ACM Symposium on Theory of Computing
Pages345-353
Number of pages9
StatePublished - 2008
Event40th Annual ACM Symposium on Theory of Computing, STOC 2008 - Victoria, BC, Canada
Duration: May 17 2008May 20 2008

Other

Other40th Annual ACM Symposium on Theory of Computing, STOC 2008
CountryCanada
CityVictoria, BC
Period5/17/085/20/08

Fingerprint

Hardness
Polynomials

Keywords

  • Approximation
  • Halfspaces
  • Hardness
  • Learning

ASJC Scopus subject areas

  • Software

Cite this

Khot, S., & Saket, R. (2008). On hardness of learning intersection of two halfspaces. In STOC'08: Proceedings of the 2008 ACM Symposium on Theory of Computing (pp. 345-353)

On hardness of learning intersection of two halfspaces. / Khot, Subhash; Saket, Rishi.

STOC'08: Proceedings of the 2008 ACM Symposium on Theory of Computing. 2008. p. 345-353.

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

Khot, S & Saket, R 2008, On hardness of learning intersection of two halfspaces. in STOC'08: Proceedings of the 2008 ACM Symposium on Theory of Computing. pp. 345-353, 40th Annual ACM Symposium on Theory of Computing, STOC 2008, Victoria, BC, Canada, 5/17/08.
Khot S, Saket R. On hardness of learning intersection of two halfspaces. In STOC'08: Proceedings of the 2008 ACM Symposium on Theory of Computing. 2008. p. 345-353
Khot, Subhash ; Saket, Rishi. / On hardness of learning intersection of two halfspaces. STOC'08: Proceedings of the 2008 ACM Symposium on Theory of Computing. 2008. pp. 345-353
@inproceedings{26c29a1ee1904ce19d7ce0c3dca10a09,
title = "On hardness of learning intersection of two halfspaces",
abstract = "We show that unless NP = RP, it is hard to (even) weakly PAC-learn intersection of two halfspaces in Rn using a hypothesis which is a, function of up to ℓ linear threshold functions for any integer ℓ. Specifically, we show that for every integer ℓ and an arbitrarily small constant ε > 0, unless NP = RP, no polynomial time algorithm can distinguish whether there is an intersection of two halfspaces that correctly classifies a given set of labeled points in Rn, or whether any function of ℓ linear threshold functions can correctly classify at most 1/2+ ε fraction of the points.",
keywords = "Approximation, Halfspaces, Hardness, Learning",
author = "Subhash Khot and Rishi Saket",
year = "2008",
language = "English (US)",
isbn = "9781605580470",
pages = "345--353",
booktitle = "STOC'08: Proceedings of the 2008 ACM Symposium on Theory of Computing",

}

TY - GEN

T1 - On hardness of learning intersection of two halfspaces

AU - Khot, Subhash

AU - Saket, Rishi

PY - 2008

Y1 - 2008

N2 - We show that unless NP = RP, it is hard to (even) weakly PAC-learn intersection of two halfspaces in Rn using a hypothesis which is a, function of up to ℓ linear threshold functions for any integer ℓ. Specifically, we show that for every integer ℓ and an arbitrarily small constant ε > 0, unless NP = RP, no polynomial time algorithm can distinguish whether there is an intersection of two halfspaces that correctly classifies a given set of labeled points in Rn, or whether any function of ℓ linear threshold functions can correctly classify at most 1/2+ ε fraction of the points.

AB - We show that unless NP = RP, it is hard to (even) weakly PAC-learn intersection of two halfspaces in Rn using a hypothesis which is a, function of up to ℓ linear threshold functions for any integer ℓ. Specifically, we show that for every integer ℓ and an arbitrarily small constant ε > 0, unless NP = RP, no polynomial time algorithm can distinguish whether there is an intersection of two halfspaces that correctly classifies a given set of labeled points in Rn, or whether any function of ℓ linear threshold functions can correctly classify at most 1/2+ ε fraction of the points.

KW - Approximation

KW - Halfspaces

KW - Hardness

KW - Learning

UR - http://www.scopus.com/inward/record.url?scp=57049098727&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=57049098727&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9781605580470

SP - 345

EP - 353

BT - STOC'08: Proceedings of the 2008 ACM Symposium on Theory of Computing

ER -