English

Complexity theoretic limitations on learning DNF's

Machine Learning 2014-11-05 v2 Computational Complexity

Abstract

Using the recently developed framework of [Daniely et al, 2014], we show that under a natural assumption on the complexity of refuting random K-SAT formulas, learning DNF formulas is hard. Furthermore, the same assumption implies the hardness of learning intersections of ω(log(n))\omega(\log(n)) halfspaces, agnostically learning conjunctions, as well as virtually all (distribution free) learning problems that were previously shown hard (under complexity assumptions).

Keywords

Cite

@article{arxiv.1404.3378,
  title  = {Complexity theoretic limitations on learning DNF's},
  author = {Amit Daniely and Shai Shalev-Shwatz},
  journal= {arXiv preprint arXiv:1404.3378},
  year   = {2014}
}

Comments

arXiv admin note: substantial text overlap with arXiv:1311.2272

R2 v1 2026-06-22T03:49:35.863Z