English

Privately Learning Decision Lists and a Differentially Private Winnow

Machine Learning 2026-02-10 v1 Cryptography and Security Machine Learning

Abstract

We give new differentially private algorithms for the classic problems of learning decision lists and large-margin halfspaces in the PAC and online models. In the PAC model, we give a computationally efficient algorithm for learning decision lists with minimal sample overhead over the best non-private algorithms. In the online model, we give a private analog of the influential Winnow algorithm for learning halfspaces with mistake bound polylogarithmic in the dimension and inverse polynomial in the margin. As an application, we describe how to privately learn decision lists in the online model, qualitatively matching state-of-the art non-private guarantees.

Keywords

Cite

@article{arxiv.2602.07370,
  title  = {Privately Learning Decision Lists and a Differentially Private Winnow},
  author = {Mark Bun and William Fang},
  journal= {arXiv preprint arXiv:2602.07370},
  year   = {2026}
}

Comments

27 pages, The 37th International Conference on Algorithmic Learning Theory

R2 v1 2026-07-01T10:25:41.507Z