Optimal Confidence Band for Kernel Gradient Flow Estimator
Yuqian Cheng [email protected], Zhuo Chen [email protected], Qian Lin [email protected] · 2026
Read Optimal Confidence Band for Kernel Gradient Flow Estimator by Yuqian Cheng [email protected], Zhuo Chen [email protected], Qian Lin [email protected] (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-2605-05768.txt.
Description
Yuqian Cheng, Zhuo Chen, and Qian Lin study uniform error bounds for a class of kernel regression methods called kernel gradient flows. Users need to know not just average error, but the worst error across the whole input range. The authors work under a standard capacity and source condition used in kernel learning theory.
They prove convergence rates for the supremum norm error of both continuous and discrete gradient flows. Those rates match the best possible minimax rates for the problem. On that base they build simultaneous confidence bands that cover the true function as a whole. The band widths shrink at rates that can come close to the best possible rate as well.
Catalog record
| CREATOR | Yuqian Cheng [email protected], Zhuo Chen [email protected], Qian Lin [email protected] |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 178 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.05768