Set-Based Training of Neural Barrier Certificates for Safety Verification of Dynamical Systems
Miriam Kranzlmüller, Lukas Koller, Tobias Ladner, Matthias Althoff · 2026
Read Set-Based Training of Neural Barrier Certificates for Safety Verification of Dynamical Systems by Miriam Kranzlmüller, Lukas Koller, Tobias Ladner, Matthias Althoff (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-2605-02526.txt.
Description
Miriam Kranzlmüller and three co-authors change how neural barrier certificates are trained for proving a system safe. A barrier certificate is a function that separates all unsafe states from all reachable ones; if it exists, safety is formally verified. Prior work trained a network, then checked it, looping until the check passed.
The new method folds the check into training itself. A set-based loss encodes every required property of the certificate in a sound way, so a loss of zero is itself a proof of validity. Training and verification merge into one procedure. Experiments show the approach scales well as system dimension grows and handles complex nonlinear dynamics naturally. For safety work on robots and other controlled systems, that merger removes a costly loop.
Catalog record
| CREATOR | Miriam Kranzlmüller, Lukas Koller, Tobias Ladner, Matthias Althoff |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 58 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.02526