Neural ODEs with Invariants by Construction
Proceedings of the AAAI Symposium Series · 06.11.2026 · 10 (1) · 695-702
DOI: 10.1609/aaaiss.v10i1.43477 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Symposium Series, online erschienen am 06.11.2026. Verfasst von Fangzhou Yu und 4 weiteren Autor:innen.
Abstract (Originalsprache)
Neural ODEs are increasingly used as continuous-time models for scientific and sensor data, but unconstrained neural ODEs can drift and violate domain invariants (e.g., conservation laws), yielding physically implausible solutions. In turn, this can compound error in long-horizon prediction and surrogate simulation. Existing solutions typically aim to enforce invari- ance by soft penalties or other forms of regularization, which can reduce overall error but do not guarantee that trajectories will not leave the constraint manifold. We introduce a frame- work that enforces invariants by construction: it treats invari- ants as first-class types and uses an LLM-driven compilation workflow to translate a generic neural ODE specification into a structure-preserving architecture whose trajectories remain on the admissible manifold in continuous time(and up to numer- ical integration error in practice). This compiler view cleanly separates what must be preserved (scientific structure) from what is learned from data (dynamics within that structure). It provides a systematic design pattern for invariant-respecting neural surrogates across scientific domains.
Zitieren
APA
Yu, F., Su, Y., Lee, R., Cheng, S., & Ramakrishnan, N. (2026). Neural ODEs with Invariants by Construction. Proceedings of the AAAI Symposium Series, 10(1), 695-702. https://doi.org/10.1609/aaaiss.v10i1.43477
Vancouver
Yu F, Su Y, Lee R, Cheng S, Ramakrishnan N. Neural ODEs with Invariants by Construction. Proceedings of the AAAI Symposium Series. 2026;10(1):695-702. doi:10.1609/aaaiss.v10i1.43477
BibTeX
@article{Yu2026,
author = {Yu, Fangzhou and Su, Yiqi and Lee, Ray and Cheng, Shenfeng and Ramakrishnan, Naren},
title = {Neural ODEs with Invariants by Construction},
journal = {Proceedings of the AAAI Symposium Series},
year = {2026},
volume = {10},
number = {1},
pages = {695--702},
doi = {10.1609/aaaiss.v10i1.43477}
}
EUVIMED