Improving the performance of Stein variational inference through extreme sparsification of physically-constrained neural network models

DOI10.1016/j.cma.2024.117359MaRDI QIDQ6641896

Reese Edward Jones, Cosmin Safta, Jan N. Fuhg, Govinda Anantha Padmanabha, Nikolaos Bouklas

Publication date: 21 November 2024

Published in: Computer Methods in Applied Mechanics and Engineering (Search for Journal in Brave)

zbMATH Keywords

neural networkprojectionuncertainty quantificationsparsificationphysical constraintsBayesian neural networkStein variational inference

Mathematics Subject Classification ID

Artificial neural networks and deep learning (68T07) Reactive materials (74A65) Numerical analysis or methods applied to Markov chains (65C40) Elastic materials (74B99) Numerical and other methods in solid mechanics (74S99) Stochastic and other probabilistic methods applied to problems in solid mechanics (74S60) Numerical approximation of high-dimensional functions; sparse grids (65D40)

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• Machine learning strategies for systems with invariance properties
• Convexity conditions and existence theorems in nonlinear elasticity
• Using the \(L\)-curve for determining optimal regularization parameters
• Bayesian learning for neural networks
• Sobolev training of thermodynamic-informed neural networks for interpretable elasto-plasticity models with level set hardening
• Learning constitutive relations using symmetric positive definite neural networks
• Data-driven tissue mechanics with polyconvex neural ordinary differential equations
• Fitting hyperelastic models to experimental data
• Modular machine learning-based elastoplasticity: generalization in the context of limited data
• Bayesian Calibration of Computer Models
• Handbook of Uncertainty Quantification
• A mechanics‐informed artificial neural network approach in data‐driven constitutive modeling
• Neural network-based multiscale modeling of finite strain magneto-elasticity with relaxed convexity criteria
• Extreme sparsification of physics-augmented neural networks for interpretable model discovery in mechanics
• A minimally invasive, efficient method for propagation of full-field uncertainty in solid dynamics
• Inverse problems: from regularization to Bayesian inference

This page was built for publication: Improving the performance of Stein variational inference through extreme sparsification of physically-constrained neural network models