Unified High-Probability Analysis of Stochastic Variance-Reduced Estimation
2026
Read Unified High-Probability Analysis of Stochastic Variance-Reduced Estimation (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-2605-15388.txt.
Introduction
Zhankun Luo, Antesh Upadhyay, M. Berk Sahin, Sang Bin Moon, Anuran Makur, and Abolfazl Hashemi unify the study of stochastic variance-reduced estimators. Past studies treat each estimator on its own and lean on averages. Their frame rests on a three-part recursion: memory hold, reset chance, and a fix for step drift. It recovers classic estimators and splits error into bias and variance parts. A new dimension-free Freedman inequality yields high-probability bounds in Euclidean and non-Euclidean settings. It also gives improved oracle costs for constrained optimization.
Catalog record
| YEAR | 2026 |
|---|---|
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 288098 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.15388