Neural DAEs: Constrained neural networks
Not on the map. Why?
Abstract (arXiv)
This article investigates the effect of explicitly adding auxiliary algebraic trajectory information to neural networks for dynamical systems. We draw inspiration from the field of differential-algebraic equations and differential equations on manifolds and implement related methods in residual neural networks, despite some fundamental scenario differences. Constraint or auxiliary information effects are incorporated through stabilization as well as projection methods, and we show when to use which method based on experiments involving simulations of multi-body pendulums and molecular dynamics scenarios. Several of our methods are easy to implement in existing code and have limited impact on training performance while giving significant boosts in terms of inference.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 16 of its 37 references that are in the map. First posted on arXiv 25 Nov 2022.
Builds on
16 of 37 references.
References · 16
- 2018Tensor field networks: Rotation- and translation-equivariant neural networks for 3D point cloudscited by 746
- 2021E(3)-Equivariant Graph Neural Networks for Data-Efficient and Accurate Interatomic Potentialscited by 708
- 2021Characterizing possible failure modes in physics-informed neural networkscited by 643
- 2020Universal Differential Equations for Scientific Machine Learningcited by 353
- 2017Stable Architectures for Deep Neural Networkscited by 341
- 2021Physics-informed neural networks with hard constraints for inverse designcited by 280
- 2017A Semantic Loss Function for Deep Learning with Symbolic Knowledgecited by 270
- 2022On Neural Differential Equationscited by 268
- 2020Integrating Scientific Knowledge with Machine Learning for Engineering and Environmental Systemscited by 164
- 2017DeePMD-kit: A deep learning package for many-body potential energy representation and molecular dynamicscited by 79
- 2016Label-Free Supervision of Neural Networks with Physics and Domain Knowledgecited by 74
- 2019Augmenting Neural Networks with First-order Logiccited by 51
4 more
- 2022Learning differentiable solvers for systems with hard constraintscited by 34
- 2022Learning continuous models for continuous physicscited by 28
- 2021Mimetic Neural Networks: A unified framework for Protein Design and Foldingcited by 7
- 2020Solving Newton's Equations of Motion with Large Timesteps using Recurrent Neural Networks based Operatorscited by 6
Built on by
Citing papers per year: 2023: 1.
1 paper.