mlsanityData to 7 Oct 2026

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

arXiv 2407.00761 ↗First posted 30 Jun 2024cs.LG cs.CEcited by 4

Not on the map. Why?

Abstract (arXiv)

Most scientific machine learning (SciML) applications of neural networks involve hundreds to thousands of parameters, and hence, uncertainty quantification for such models is plagued by the curse of dimensionality. Using physical applications, we show that L0 sparsification prior to Stein variational gradient descent (L0+SVGD) is a more robust and efficient means of uncertainty quantification, in terms of computational cost and performance than the direct application of SGVD or projected SGVD methods. Specifically, L0+SVGD demonstrates superior resilience to noise, the ability to perform well in extrapolated regions, and a faster convergence rate to an optimal solution.

Key facts. Cited by 4 papers in the map (data through 7 Oct 2026). Builds on 24 of its 55 references that are in the map. First posted on arXiv 30 Jun 2024.

Builds on

24 of 55 references.

References · 24

  1. 2016Variational Inference: A Review for Statisticianscited by 1,466
  2. 2018Tensor field networks: Rotation- and translation-equivariant neural networks for 3D point cloudscited by 735
  3. 2017Learning Sparse Neural Networks through L0 Regularizationcited by 571
  4. 2016Stein Variational Gradient Descent: A General Purpose Bayesian Inference Algorithmcited by 498
  5. 2016Input Convex Neural Networkscited by 407
  6. 2017Advances in Variational Inferencecited by 236
  7. 2017A Conceptual Introduction to Hamiltonian Monte Carlocited by 207
  8. 2017Stein Variational Gradient Descent as Gradient Flowcited by 149
  9. 2017Bayesian Compression for Deep Learningcited by 116
  10. 2020Bayesian Bits: Unifying Quantization and Pruningcited by 93
  11. 2021Polyconvex anisotropic hyperelasticity with neural networkscited by 54
  12. 2020Projected Stein Variational Gradient Descentcited by 33
12 more
  1. 2021Data-driven Tissue Mechanics with Polyconvex Neural Ordinary Differential Equationscited by 28
  2. 2019Projected Stein Variational Newton: A Fast and Scalable Bayesian Inference Method in High Dimensionscited by 21
  3. 2022Modular machine learning-based elastoplasticity: generalization in the context of limited datacited by 16
  4. 2023Extreme sparsification of physics-augmented neural networks for interpretable model discovery in mechanicscited by 16
  5. 2018Stein Variational Gradient Descent as Moment Matchingcited by 15
  6. 2023Neural network-based multiscale modeling of finite strain magneto-elasticity with relaxed convexity criteriacited by 14
  7. 2021Equivariant geometric learning for digital rock physics: estimating formation factor and effective permeability tensors from Morse graphcited by 11
  8. 2023Stress representations for tensor basis neural networks: alternative formulations to Finger-Rivlin-Ericksencited by 9
  9. 2022On the detrimental effect of invariances in the likelihood for variational inferencecited by 8
  10. 2022A stochastic Stein Variational Newton methodcited by 4
  11. 2023Bridging scales with Machine Learning: From first principles statistical mechanics to continuum phase field computations to study order disorder transitions in LixCoO2cited by 3
  12. 2023Accurate Data-Driven Surrogates of Dynamical Systems for Forward Propagation of Uncertaintycited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 2; 2026: 2.

4 papers.

  1. 2025Uncertainty quantification of neural network models of evolving processes via Langevin samplingcited by 2
  2. 2025Fully data-driven inverse hyperelasticity with hyper-network neural ODE fieldscited by 0
  3. 2026CANN-EUCLID: unsupervised constitutive artificial neural network model discovery from full-field datacited by 0
  4. 2026Interval and fuzzy physics-augmented neural networks (iPANN and fPANN) for uncertainty quantification and propagation in constitutive modelingcited by 0