Salvato in:
Dettagli Bibliografici
Autori principali: Şahin, Yakup Emre, Kilbertus, Niki, Becker, Sören
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2510.23295
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866908614592036864
author Şahin, Yakup Emre
Kilbertus, Niki
Becker, Sören
author_facet Şahin, Yakup Emre
Kilbertus, Niki
Becker, Sören
contents We introduce MIO, a transformer-based model for inferring symbolic ordinary differential equations (ODEs) from multiple observed trajectories of a dynamical system. By combining multiple instance learning with transformer-based symbolic regression, the model effectively leverages repeated observations of the same system to learn more generalizable representations of the underlying dynamics. We investigate different instance aggregation strategies and show that even simple mean aggregation can substantially boost performance. MIO is evaluated on systems ranging from one to four dimensions and under varying noise levels, consistently outperforming existing baselines.
format Preprint
id arxiv_https___arxiv_org_abs_2510_23295
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Predicting symbolic ODEs from multiple trajectories
Şahin, Yakup Emre
Kilbertus, Niki
Becker, Sören
Machine Learning
We introduce MIO, a transformer-based model for inferring symbolic ordinary differential equations (ODEs) from multiple observed trajectories of a dynamical system. By combining multiple instance learning with transformer-based symbolic regression, the model effectively leverages repeated observations of the same system to learn more generalizable representations of the underlying dynamics. We investigate different instance aggregation strategies and show that even simple mean aggregation can substantially boost performance. MIO is evaluated on systems ranging from one to four dimensions and under varying noise levels, consistently outperforming existing baselines.
title Predicting symbolic ODEs from multiple trajectories
topic Machine Learning
url https://arxiv.org/abs/2510.23295