On Small deviations of Gaussian processes using majorizing measures
Michel Weber

TL;DR
This paper explores small deviations in Gaussian processes, demonstrating that processes with similar entropy can have vastly different behaviors, and introduces a sharp lower bound using majorizing measures.
Contribution
It provides new examples of Gaussian processes with identical entropy but different small deviations and establishes a general, sharp lower bound via majorizing measures.
Findings
Examples of Gaussian processes with same entropy but different small deviations
A sharp lower bound for small deviations using majorizing measures
Application of bounds to Gaussian sequences and ultrametric processes
Abstract
We give two examples of periodic Gaussian processes, having entropy numbers of exactly same order but radically different small deviations. Our construction is based on classical Knopp's result yielding of existence of continuous nowhere differentiable functions, and more precisely on Loud's functions. We also obtain a general lower bound for small deviations using the majorizing measure method. We show on examples that our bound is sharp. We also apply it to Gaussian independent sequences and to the generic class of ultrametric Gaussian processes. 60F15, 60G50 ; Secondary: 60F05
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
