Model discovery of compartmental models with graph-supported neural networks

DOI10.1016/J.AMC.2023.128392MaRDI QIDQ6090296zbMATH OpenOpenAlexFDO

Authors Fernando Javier Aguilar-Canto, Carlos Brito-Loeza, Hiram Calvo

Publication date 14 November 2023

Published in Applied Mathematics and Computation (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1016/j.amc.2023.128392

zbMATH Keywords

compartmental modelsgraph-supported neural networksmodel discoveryneural network approximationordinary differential equations

Mathematics Subject Classification ID

Qualitative theory for ordinary differential equations (34Cxx) Inference from stochastic processes (62Mxx) Artificial intelligence (68Txx)

Recommendations

• Deep learning of biological models from data: applications to ODE models
• KAN-ODEs: Kolmogorov-Arnold network ordinary differential equations for learning dynamical systems and hidden physics
• Structure-preserving recurrent neural networks for a class of Birkhoffian systems
• Modeling of complex dynamic systems using differential neural networks with the incorporation of a priori knowledge

Cites work

• Approximation by superpositions of a sigmoidal function
• Automated reverse engineering of nonlinear dynamical systems
• Bayesian physics informed neural networks for real-world nonlinear dynamical systems
• Causal inference in statistics. A primer
• Characterization of a two-parameter mixed-mode electrochemical behavior regime using neural networks
• Deep neural networks motivated by partial differential equations
• GINNs: graph-informed neural networks for multiscale physics
• Global dynamics in a Beddington-DeAngelis prey-predator model with density dependent death rate of predator
• Global stability for SIRS epidemic models with general incidence rate and transfer from infectious to susceptible
• Nonlinear Laplacian spectral analysis for time series with intermittency and low-frequency variability
• PDE-Net 2.0: learning PDEs from data with a numeric-symbolic hybrid deep network
• Physics-informed graph neural Galerkin networks: a unified framework for solving PDE-governed forward and inverse problems
• Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Solving many-electron Schrödinger equation using deep neural networks

This page was built for publication: Model discovery of compartmental models with graph-supported neural networks

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6090296)