On the Precise Asymptotics of Universal Inference
Kenta Takatsu · 2025
Read On the Precise Asymptotics of Universal Inference by Kenta Takatsu (2025) 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-2503-14717.txt.
Description
Takatsu studies why some confidence sets are far wider than they need to be in twenty twenty-five. A valid method should cover the true value at its stated rate. Some methods cover it almost always, which means their sets are needlessly wide. The paper uses universal inference as its running example under model misspecification.
The author traces the source of this asymptotic conservativeness. He proposes a general fix based on studentization and bias correction. The fixed method reaches exact coverage at the stated level in the long run. It works even when the model is misspecified, under a condition on estimation errors. That condition resembles double robustness in semiparametric theory. The work aims to make confidence sets both valid and sharp.
Catalog record
| CREATOR | Kenta Takatsu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 107 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-14717