Complexity Guarantees for Zeroth-order Methods via Exponentially-shifted Gaussian Smoothing: Mitigating Dimension-dependence and Incorporating Decision-dependence
MINGRUI WANG, PRAKASH CHAKRABORTY, UDAY V. SHANBHAG · 2026
Read Complexity Guarantees for Zeroth-order Methods via Exponentially-shifted Gaussian Smoothing: Mitigating Dimension-dependence and Incorporating Decision-dependence by MINGRUI WANG, PRAKASH CHAKRABORTY, UDAY V. SHANBHAG (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-15525.txt.
Description
Wang, Chakraborty, and Shanbhag introduce exponentially-shifted Gaussian smoothing for zeroth-order optimization, whose estimate moments grow linearly rather than quadratically with dimension. The estimate stretches to decision-linked cases with densities known or unknown in closed form, still linear in dimension. Their method covers nonsmooth strongly convex, convex, and non-convex cases with better dimension scaling in iteration counts and new high-chance guarantees.
Catalog record
| CREATOR | MINGRUI WANG, PRAKASH CHAKRABORTY, UDAY V. SHANBHAG |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 153235 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.15525