Adaptive weighting of Bayesian physics informed neural networks for multitask and multiscale forward and inverse problems

DOI10.1016/j.jcp.2023.112342arXiv2302.12697OpenAlexW4383558850MaRDI QIDQ6095075

Suryanarayana Maddu, Sarah Perez, Philippe Poncet, Ivo F. Sbalzarini

Publication date: 27 November 2023

Published in: Journal of Computational Physics (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/2302.12697

zbMATH Keywords

artificial intelligenceuncertainty quantificationHamiltonian Monte CarloBayesian physics-informed neural networksadaptive weight learningmulti-objective training

Mathematics Subject Classification ID

Parametric inference (62Fxx) Artificial intelligence (68Txx) Probabilistic methods, stochastic differential equations (65Cxx)

Cites Work

• Unnamed Item
• The No-U-Turn Sampler: Adaptively Setting Path Lengths in Hamiltonian Monte Carlo
• Discovering governing equations from data by sparse identification of nonlinear dynamical systems
• Using DC PSE operator discretization in Eulerian meshless collocation methods improves their robustness in complex geometries
• Sobolev training of thermodynamic-informed neural networks for interpretable elasto-plasticity models with level set hardening
• Optimally weighted loss functions for solving PDEs with neural networks
• A velocity-vorticity method for highly viscous 3D flows with application to digital rock physics
• B-PINNs: Bayesian physics-informed neural networks for forward and inverse PDE problems with noisy data
• Multi-fidelity Bayesian neural networks: algorithms and applications
• Physics-informed semantic inpainting: application to geostatistical modeling
• When and why PINNs fail to train: a neural tangent kernel perspective
• Multi-variance replica exchange SGMCMC for inverse and forward problems via Bayesian PINN
• Surrogate modeling for fluid flows based on physics-constrained deep learning without simulation data
• Adaptive multi-fidelity polynomial chaos approach to Bayesian inference in inverse problems
• Adversarial uncertainty quantification in physics-informed neural networks
• Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• On the geometric ergodicity of Hamiltonian Monte Carlo
• The geometric foundations of Hamiltonian Monte Carlo
• Bayesian physics informed neural networks for real-world nonlinear dynamical systems
• Uncertainty quantification in scientific machine learning: methods, metrics, and comparisons
• Spatially distributed control for optimal drag reduction of the flow past a circular cylinder
• Riemann Manifold Langevin and Hamiltonian Monte Carlo Methods
• Learning partial differential equations via data discovery and sparse optimization
• Understanding and Mitigating Gradient Flow Pathologies in Physics-Informed Neural Networks
• Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations
• An Adaptive Surrogate Modeling Based on Deep Neural Networks for Large-Scale Bayesian Inverse Problems
• Contribution to the Discussion of the Paper ‘Geodesic Monte Carlo on Embedded Manifolds’

This page was built for publication: Adaptive weighting of Bayesian physics informed neural networks for multitask and multiscale forward and inverse problems