English

Optimal Learning Rates for Regularized Least-Squares with a Fourier Capacity Condition

Statistics Theory 2023-09-19 v4 Functional Analysis Machine Learning Statistics Theory

Abstract

We derive minimax adaptive rates for a new, broad class of Tikhonov-regularized learning problems in Hilbert scales under general source conditions. Our analysis does not require the regression function to be contained in the hypothesis class, and most notably does not employ the conventional \textit{a priori} assumptions on kernel eigendecay. Using the theory of interpolation, we demonstrate that the spectrum of the Mercer operator can be inferred in the presence of ``tight'' L(X)L^{\infty}(\mathcal{X}) embeddings of suitable Hilbert scales. Our analysis utilizes a new Fourier isocapacitary condition, which captures the interplay of the kernel Dirichlet capacities and small ball probabilities via the optimal Hilbert scale function.

Keywords

Cite

@article{arxiv.2204.07856,
  title  = {Optimal Learning Rates for Regularized Least-Squares with a Fourier Capacity Condition},
  author = {Prem Talwai and David Simchi-Levi},
  journal= {arXiv preprint arXiv:2204.07856},
  year   = {2023}
}

Comments

The first version of this paper contained an error -- refer to current version

R2 v1 2026-06-24T10:49:59.645Z