arXiv subjects

Branwen Snelling

Publications and source records attributed to Branwen Snelling.

2 recordsLinked to original sources

arXiv2026

PEtab SciML: an exchange format for specifying and training dynamic scientific machine learning models

Summary: Dynamic scientific machine learning (SciML) models that combine mechanistic ordinary differential equations (ODEs) with machine learning (ML) components have applications ranging from learning unknown biological processes to integrating auxiliary data modalities into dynamic modelling. To enable reproducible and efficient SciML training, we introduce PEtab SciML, an interoperable data format for specifying parameter estimation problems in which mechanistic and ML model parameters are jointly estimated from time series data. PEtab SciML supports several ML ODE hybridization patterns in realistic problem setups. It is accompanied by a reference Python library and downstream modelling support in Python/JAX and Julia, provided by AMICI and PEtab$.$jl, respectively, and a collection of real data benchmarks. Availability and implementation: PEtab SciML is available on GitHub (https://github.com/PEtab-dev/petab_sciml). The reference Python package is installable from PyPI and is continuously tested and supported on Linux, macOS, and Windows.

q-bio.QM↗

arXiv2026

Curriculum Multiple Shooting for Robust Training of Neural and Universal Differential Equations

Neural ordinary differential equations (NODEs) and universal differential equations (UDEs) provide flexible and popular frameworks for learning interpretable dynamical systems from noisy time-series data. However, training these models remains challenging, and versatile methods that robustly handle sparse and noisy data as well as partially observed models are lacking. To address this, we introduce curriculum multiple shooting (CMS), a general-purpose training strategy for fitting ordinary differential equation (ODE) models to time-series data by integrating curriculum learning with multiple shooting. Across twelve benchmarks spanning simulated and real data, and covering NODEs, UDEs, and mechanistic ODEs, CMS accelerates and stabilises training convergence, outperforms state-of-the-art training strategies, and ranks among the best methods in generalisation. Finally, we discuss possible explanations for the strong performance of CMS in light of contemporary theories of what makes training on time-series challenging.

q-bio.QM↗