Distributionally Robust PAC-Bayesian Control
Domagoj Herceg, Duarte Antunes · 2026
Read Distributionally Robust PAC-Bayesian Control by Domagoj Herceg, Duarte Antunes (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-2604-10588.txt.
Description
Herceg and Antunes build a safety-certification framework for controllers learned from data. It covers cases where training and real deployment differ and losses need not stay bounded. They combine PAC-Bayes generalization theory with distributionally robust optimization based on Wasserstein distance. From this they derive a loss proxy and a bound on the damage from distribution shift, both tied to the closed-loop map. For linear time-invariant systems the result is a tractable optimization scheme with high-probability safety certificates for real deployment.
Catalog record
| CREATOR | Domagoj Herceg, Duarte Antunes |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 38851 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.10588