arXiv Open Access 2026

Privately Learning Decision Lists and a Differentially Private Winnow

Mark Bun William Fang

Lihat Sumber

Abstrak

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.

Topik & Kata Kunci

cs.LG cs.CR stat.ML

Penulis (2)

Mark Bun

William Fang

Format Sitasi

APA MLA BibTeX

Bun, M., Fang, W. (2026). Privately Learning Decision Lists and a Differentially Private Winnow. https://arxiv.org/abs/2602.07370

Akses Cepat

Lihat di Sumber

Informasi Jurnal

Tahun Terbit: 2026
Bahasa: en
Sumber Database: arXiv
Akses: Open Access ✓