Mixed precision training of neural ODEs

DOI10.1137/25M1821405MaRDI QIDQ7410853zbMATH OpenFDO

Authors Elena Celledoni, Brynjulf Owren, Lars Ruthotto, Nicole Tianjiao Yang

Publication date 2026

Published in SIAM Journal on Scientific Computing (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1137/25M1821405

zbMATH Keywords

deep learningmixed precision computingneural ODEs

Mathematics Subject Classification ID

Roundoff error (65G50) Multistep, Runge-Kutta and extrapolation methods for ordinary differential equations (65L06) Artificial neural networks and deep learning (68T07)

Cites work

• A proposal on machine learning via dynamical systems
• Accuracy and Stability of Numerical Algorithms
• Deep neural networks motivated by partial differential equations
• Introduction to numerical analysis. Transl. from the German by R. Bartels, W. Gautschi, and C. Witzgall.
• Mixed precision algorithms in numerical linear algebra
• Mixed precision block fused multiply-add: error analysis and application to GPU tensor cores
• Mixed-precision explicit stabilized Runge-Kutta methods for single- and multi-scale differential equations
• Perturbed Runge-Kutta methods for mixed precision applications
• Speeding up and reducing memory usage for scientific machine learning via mixed precision
• Stable architectures for deep neural networks

This page was built for publication: Mixed precision training of neural ODEs

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7410853)