English

Dimension-free comparison estimates for suprema of some canonical processes

Probability 2024-11-06 v2 Statistics Theory Statistics Theory

Abstract

We obtain error approximation bounds between expected suprema of canonical processes that are generated by random vectors with independent coordinates and expected suprema of Gaussian processes. In particular, we obtain a sharper proximity estimate for Rademacher and Gaussian complexities. Our estimates are dimension-free, and depend only on the geometric parameters and the numerical complexity of the underlying index set.

Keywords

Cite

@article{arxiv.2312.14308,
  title  = {Dimension-free comparison estimates for suprema of some canonical processes},
  author = {Shivam Sharma},
  journal= {arXiv preprint arXiv:2312.14308},
  year   = {2024}
}

Comments

To appear in High Dimensional Probability Volume 10