Pith. sign in

REVIEW 4 major objections 5 minor 38 references

Physics-informed neural networks (PINNs) for numerical model error approximation and superresolution

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Physics-informed loss terms let a neural network approximate the nodal difference between coarse and fine finite-element solutions and output the fine mesh's displacement field, beating a purely data-driven baseline.

desk verdict The paper's core ablation is invalid: Eq. (7) is algebraically identical to beta_1 times Eq. (6), so the claimed physics-informed improvement is just loss reweighting. read the letter →

arxiv 2411.09728 v1 pith:NMAYKMZL submitted 2024-11-14 cs.LG cs.NAmath.NAstat.CO

Bozhou Zhuang , Sashank Rana , Brandon Jones , Danny Smyl This is my paper · ORCID
classification cs.LGcs.NAmath.NAstat.CO
keywords physics-informedneuralnetworksmodelerrorapproximationfiniteelementanalysissuperresolutionQ4/Q8discretizationdisplacementfieldsuncertaintyquantificationelasticplatewithopening
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to show that a neural network can explicitly approximate numerical model errors in finite element analysis—the gap between a coarse four-node quadrilateral (Q4) mesh solution and a fine eight-node quadrilateral (Q8) mesh solution—rather than only supplying a scalar error bound or a corrected operator. The proposed physics-informed neural network (PINN) takes the Q4 displacement field as input and outputs both the model-error field at Q4 nodes and the Q8 displacement field, effectively combining error approximation with superresolution. Trained on 9,000 randomized samples of a two-dimensional elastic plate with a central opening and tested on 1,000 held-out samples, the network reproduces the error fields closely, with mean absolute differences below $4.0\times10^{-6}$ m in most areas. The central comparison is an ablation showing that adding a displacement-consistency physics loss reduces mean testing error by 7.3% relative to the purely data-driven setup, which is the evidence for the paper's claim that physics-informed loss functions let networks surpass data-only approximation of model errors.

What carries the argument

The central object is a two-branch physics-informed neural network whose input is the Q4 nodal displacement vector (1,722 entries). One branch outputs predicted model error at Q4 nodes (1,722 entries); the other outputs a predicted Q8 displacement field (19,682 entries), which is the superresolution branch. Training minimizes a weighted total loss $\mathcal{L}_{\text{total}} = \mathcal{L}_{\text{error}} + \beta_1 \mathcal{L}_u + \beta_2 \mathcal{L}_{\text{super}}$, where $\mathcal{L}_u$ is the displacement-consistency term that penalizes deviations of the corrected Q4 field from the Q8 solution sampled at Q4 nodes, and $\mathcal{L}_{\text{super}}$ directly supervises the Q8 output. The $\mathcal{L}_u$ term is the load-bearing physics-informed component: the ablation shows it reduces both training (12.0%) and testing (7.3%) model-error loss, while $\mathcal{L}_{\text{super}}$ adds the higher-resolution output. Dropout is kept active at test time to approximate predictive uncertainty.

What would settle it

Take several test samples and compare the Q8 solution against a substantially finer reference mesh (for example, 40,000 or more Q8 nodes) or against an analytical solution. If the per-node difference between Q8 and that reference is comparable to or larger than the PINN's predicted model error, then correcting Q4 toward Q8 cannot reduce the true error, and the claim that the network approximates numerical model error would be falsified.

Watch

Extended reading notes

Core claim

The central claim is that the learned error approximator $A(\Theta)$ in $\tilde{u}_H \approx u_R + A(\Theta)$ captures the numerical model error $e = u_H - u_R$ closely enough that correcting a Q4 solution moves it toward Q8 accuracy, and that the same network can directly emit the Q8 displacement field. The support comes from a two-branch network trained with three loss terms: $\mathcal{L}_{\text{error}}$ for the nodal error, $\mathcal{L}_u$ enforcing that error-plus-Q4-input match the Q8 displacement sampled on the Q4 mesh, and $\mathcal{L}_{\text{super}}$ enforcing that the second branch match the full Q8 displacement. Removing $\mathcal{L}_u$ raises mean testing loss by 7.3%, and removing both physics terms raises it further, so the physics constraints contribute beyond the data-driven mapping. Dropout-based uncertainty places prediction spread at $2{-}2.5\times10^{-6}$ m, well below the model-error magnitudes, while the superresolution branch produces Q8 fields directly from Q4 inputs.

Load-bearing premise

The load-bearing premise is that the Q8 solution at 9,841 nodes is the true higher-order answer; if that mesh is itself not converged, the network is learning the difference between two imperfect meshes rather than the actual numerical model error.

Editorial extensions

If this is right

  • Coarse Q4 solutions can be corrected to Q8-level nodal accuracy with a single forward pass, giving engineers explicit spatial error maps instead of scalar error bounds.
  • Adding the displacement-consistency physics loss ($\mathcal{L}_u$) is shown to cut testing model-error loss by 7.3% over the data-only baseline, so physics-informed regularization provides a measurable accuracy gain.
  • The dual output makes the network a superresolver in both mesh fineness ($h$) and interpolation order ($p$), producing Q8 displacement fields from Q4 inputs.
  • Dropout uncertainty estimates in the $2{-}2.5\times10^{-6}$ m range give a practical confidence measure for the predicted error field.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A direct testable extension is to hold out a third, even finer reference solution (or an analytical solution) and ask whether the corrected Q4 field moves toward that reference; if it does not, the learned quantity is mesh-to-mesh difference rather than true model error.
  • The correction formula $\tilde{u}_H \approx u_R + A(\Theta)$ could be applied recursively: feed the corrected field back as input to approximate the error at the next refinement level, enabling multi-level error correction without re-running the fine solver.
  • Because the input is the raw Q4 nodal field, the same two-branch architecture could be retrained for other mesh pairs (for example, linear-to-quadratic or triangular-to-quadrilateral elements) if corresponding randomized datasets are generated.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The manuscript proposes a neural-network-based explicit model error approximator for finite element solutions. The network takes reduced-order (Q4) displacement fields as input and outputs both a predicted model error field (the Q4-to-Q8 difference) and a Q8 displacement field for superresolution. The training loss combines an error term, a displacement loss, and a superresolution loss, with the last two described as physics-informed. Using a synthetic dataset of 10,000 elastic plate simulations, the authors compare three loss configurations and report that including the displacement loss improves testing accuracy over purely data-driven training. The paper also presents dropout-based uncertainty estimates and qualitative superresolution results.

Significance. If correct, explicit machine-learned error approximators that correct coarse finite element solutions toward finer ones would be practically useful, and the paper's dataset-generation pipeline is a reasonable test bed. The basic error-prediction results look plausible: prediction-difference histograms are centered near zero and mean absolute differences are below 4e-6 m. However, the central comparative claim is not supported: the displacement loss is algebraically identical to the error loss, so the reported 'physics-informed' gain is a reweighting artifact. The paper also does not verify that the Q8 solution is a converged ground truth and does not provide code. These issues undermine the stated novelty and the title/abstract claims.

major comments (4)
  1. [Loss Functions and Model Training] Equation (7) defines Lu = (β1/N) Σ ||(e^A_n + u_R,n) − u_(H,Ω_Q4),n||_1. Since the true model error in Eq. (1) is e^d_n = u_(H,Ω_Q4),n − u_R,n, each term of Lu equals ||e^A_n − e^d_n||_1, making Lu = β1·Lerror. Consequently the total loss in Case 2 is (1 + β1)·Lerror, a scaled version of the Case 3 loss. A learnable β1 cannot inject independent physical information; it only rescales the gradient magnitude. The reported improvement between Cases 2 and 3 (testing loss 1.14 vs 1.23 × 10^-6 m) is smaller than the reported standard deviations (2.90 and 2.30 × 10^-6 m) and therefore cannot be attributed to physics. This invalidates the headline claim that physics-informed loss functions outperform a purely data-driven approach.
  2. [Numerical Simulation and Dataset Generation] The Q8 discretization is designated as the 'higher-order model' and its solution is used as ground truth in Eq. (1), but no convergence study is provided. Q4 and Q8 differ simultaneously in polynomial order and mesh density (861 vs 9,841 nodes). Without demonstrating that the Q8 solution is effectively converged with respect to h- and p-refinement, the learned target is only a difference between two imperfect discretizations, and correcting Q4 toward Q8 does not necessarily reduce the actual numerical error. This premise is load-bearing because the entire supervised objective is defined by this difference.
  3. [Loss Functions and Model Training] The term 'physics-informed' is used for losses that are not physics-based. Lsuper in Eq. (8) is a standard supervised loss comparing a network output to Q8 nodal values; it imposes no PDE, equilibrium, or constitutive constraint. Lu, as shown above, reduces to the data-fidelity error loss. No physical residual is used anywhere in the network. Thus the comparison in Table 1 does not isolate any physics-informed mechanism, and the description of the method as a PINN is misleading.
  4. [Superresolution] The claim that the method achieves superresolution in both the h and p senses is not supported. The network maps Q4 nodal displacements to Q8 nodal displacements, which is an h-type upsampling (more nodes). No mechanism changes the element interpolating polynomial basis, so the claimed p-refinement novelty is not demonstrated. The superresolution results are also only shown qualitatively for one sample, without a quantitative error metric on the testing set.
minor comments (5)
  1. [Numerical Simulation] There are duplicated phrases in the text: 'conducted using using' and 'outputs from the from the numerical simulation' should be corrected.
  2. [Table 1] Table 1 has formatting issues ('1 .38') and does not specify whether the reported standard deviations are over epochs, over seeds, or over testing samples; the testing standard deviations are larger than the mean differences used to support the physics-informed gain, so the uncertainty should be quantified more carefully.
  3. [Formulation] The notation for the corrected solution, written as 'e_u_H', is undefined and appears to be a typo; it should be clearly distinguished from the true higher-order solution u_H.
  4. [Data Availability Statement] The data availability statement says data are available upon reasonable request, but no code or trained model is provided; releasing code would substantially improve reproducibility, especially given the difficulty of reproducing the exact loss-weighting scheme.
  5. [Loss Functions] The learned values of the loss coefficients β1 and β2 are never reported, so the reader cannot evaluate the effective weighting of the loss terms or compare the results across configurations.

Circularity Check

1 steps flagged · score 7.0 of 10

Eq. (7) Lu is algebraically β1 times Eq. (6) Lerror, so the claimed physics-informed improvement over data-driven training is a loss-reweighting artifact, not evidence of physics.

  1. self definitional [Loss Functions and Model Training (Section 5), Eqs. (6)-(7) with Eq. (1)]
    "Lu= 𝛽1 𝑁 𝑁∑︁ 𝑛=1 ∥(𝑒A 𝑛(Θ)+ 𝑢𝑅,𝑛)− 𝑢(𝐻,Ω𝑄4),𝑛∥1 (7) ... Lerror= 1 𝑁 𝑁∑︁ 𝑛=1 ||𝑒A 𝑛(Θ)− 𝑒𝑑 𝑛|| 1 (6)"

    By Eq. (1) and the dataset definition, the true error at coincident Q4 nodes is e^d_n = u_(H,ΩQ4),n - u_R,n. Substituting into Eq. (7), each term (e^A_n + u_R,n) - u_(H,ΩQ4),n equals e^A_n - e^d_n, so Lu = β1 Lerror exactly. Thus Case 2's total loss Lerror + Lu = (1+β1)Lerror is the same objective as Case 3's purely data-driven Lerror up to a positive scalar; the learnable β1 only rescales gradients and cannot encode independent physics. The reported improvement of Case 2 over Case 3 (1.14e-6 vs 1.23e-6, with standard deviations 2.90e-6 and 2.30e-6) is therefore a loss-weighting artifact, and the headline claim that physics-informed losses surpass data-driven training reduces by construction to reweighting the data-fidelity term.

full rationale

The central comparative claim is circular: the displacement 'physics' loss is definitionally the same as the error loss, so the evidence that physics-informed training outperforms purely data-driven training reduces to a rescaling of the same data-fidelity objective. The error-approximation and superresolution tasks themselves remain legitimate supervised regression problems, and Lsuper is an independent supervised output; self-citations (e.g., Smyl et al. 2021 for the input choice, Zhuang et al. 2024 for dropout uncertainty) are not load-bearing. The circularity is confined to the physics-vs-data-driven comparison, but that comparison is the paper's headline claim, warranting a score of 7 rather than lower.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central claim rests on treating Q8 as exact, on the learnability of A, and on the assertion that Lu and Lsuper provide independent physical constraints. The first two are standard but unverified here; the third is false because Lu = beta_1 * Lerror by Eqs. (6) and (7).

free parameters (4)
  • Learnable loss coefficients beta_1 and beta_2 = not reported
    Used to weight Lu and Lsuper in Eqs. (7) and (8); tuned during training; no final values are given.
  • PReLU negative slopes = not reported
    Learnable per-channel slopes in the PReLU activation layers.
  • Neural network weights and biases = not reported, no checkpoint
    All parameters of the Linear layers are fitted to the 9,000 training samples; no model artifact is released.
  • Training hyperparameters = LR=1e-5, gamma=0.99, batch=32, dropout=0.1
    Chosen by hand, not fitted to data, but they affect the reported results and are not justified by a sensitivity study.
assumptions (4)
  • domain assumption The Q8 finite element solution is a suitable higher-order reference for defining model errors.
    Eq. (1) defines e = u_H - u_R with Q8 as u_H; no mesh convergence or reference solution check is reported in Section 3.
  • domain assumption The reduced-order displacement field u_R contains enough information for A(u_R) to predict e for unseen material and loading samples.
    Section 2 sets Theta = u_R in Eq. (2); generalization across randomized fields is assumed without a theoretical or transferability study.
  • domain assumption Standard linear elasticity governs the plate and the generated FEM data are accurate realizations of that model.
    Dataset generation in Section 3 assumes an elastic plate with randomized modulus fields; no experimental validation is attempted.
  • domain assumption Training and test samples are i.i.d. draws from the same distribution of forces and modulus fields.
    The 10,000 samples are generated by randomizing loads and material fields, and the split is random, but no distribution shift analysis is given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Physics-informed neural networks (PINNs) for numerical model error approximation and superresolution." pith.science (2026). https://pith.science/paper/NMAYKMZL

@misc{pith2026241109728,
  author       = {Pith},
  title        = {Pith review of: Physics-informed neural networks (PINNs) for numerical model error approximation and superresolution},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NMAYKMZL}},
  note         = {Machine review of arXiv:2411.09728}
}
read the original abstract

Numerical modeling errors are unavoidable in finite element analysis. The presence of model errors inherently reflects both model accuracy and uncertainty. To date there have been few methods for explicitly quantifying errors at points of interest (e.g. at finite element nodes). The lack of explicit model error approximators has been addressed recently with the emergence of machine learning (ML), which closes the loop between numerical model features/solutions and explicit model error approximations. In this paper, we propose physics-informed neural networks (PINNs) for simultaneous numerical model error approximation and superresolution. To test our approach, numerical data was generated using finite element simulations on a two-dimensional elastic plate with a central opening. Four- and eight-node quadrilateral elements were used in the discretization to represent the reduced-order and higher-order models, respectively. It was found that the developed PINNs effectively predict model errors in both x and y displacement fields with small differences between predictions and ground truth. Our findings demonstrate that the integration of physics-informed loss functions enables neural networks (NNs) to surpass a purely data-driven approach for approximating model errors.

Figures

Figures reproduced from arXiv: 2411.09728 by the authors.

Figure 1
Figure 1. Illustration depicting (a) a classic symmetric elastic plate stretching problem solved using 1,800 finite element 4- and 16-node quadrilateral discretizations, denoted with ’4’ and ’16’ superscripts, respectively; (b) finite element displacement solutions separated into x and y fields for coincidental nodes; (c) x- and y￾displacement error heat maps; (d) the same displacement errors depicted via line graphs; and (e)… view at source ↗
Figure 2
Figure 2. Displacement fields of lower-order and higher-order models and model error in the generated dataset. 20 Zhuang et al., November 18, 2024 [PITH_FULL_IMAGE:figures/full_fig_p020_2.png] view at source ↗
Figure 3
Figure 3. [PITH_FULL_IMAGE:figures/full_fig_p021_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Distributions of prediction differences of all nodes in the testing set. (a) Mode error difference with Q4 mesh. This histogram is obtained by subtracting the true model error from the predicted model error at all nodes, and (b) displacement field difference with Q8 me…
Figure 5
Figure 5. Figure 5: Comparison of model errors for a single test sample between PINN prediction and ground truth. 23 Zhuang et al., November 18, 2024 [PITH_FULL_IMAGE:figures/full_fig_p023_5.png]
Figure 6
Figure 6. Figure 6: Absolute difference between predicted model errors and ground truth in the overall testing set. 24 Zhuang et al., November 18, 2024 [PITH_FULL_IMAGE:figures/full_fig_p024_6.png]
Figure 7
Figure 7. Figure 7: Nodal comparisons of model errors between PINN prediction and ground truth of the same sample in [PITH_FULL_IMAGE:figures/full_fig_p025_7.png]
Figure 8
Figure 8. Figure 8: Nodal comparisons of Q8 displacement between PINN prediction and ground truth of the same sample in [PITH_FULL_IMAGE:figures/full_fig_p026_8.png]
Figure 9
Figure 9. Figure 9: Loss curves of cases listed in [PITH_FULL_IMAGE:figures/full_fig_p027_9.png]
Figure 10
Figure 10. Figure 10: Prediction uncertainty of the same testing sample in [PITH_FULL_IMAGE:figures/full_fig_p028_10.png]
Figure 11
Figure 11. Figure 11: Superresolution of the 𝑥-displacement field with Q8 discretization using the same testing sample in [PITH_FULL_IMAGE:figures/full_fig_p029_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 34 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION article output.bibitem format.authors "author" output.check author format.key output output.year.check new.block format.title "title" output.check after.quote 'output.state := crossref missing journal emphasize "journal" output.check organization output format.vol.num.pages output format.article.cros...

  2. [2]

    Arcaro, A., Zhuang, B., Gencturk, B., and Ghanem, R. (2024). ``Damage detection and localization in sealed spent nuclear fuel dry storage canisters using multi-task machine learning classifiers.''\ Reliability Engineering & System Safety , 252, 110446

    work page 2024

  3. [3]

    Arridge, S., Hauptmann, A., and Korolev, Y. (2023). ``Inverse problems with learned forward operators.''\ arXiv preprint arXiv:2311.12528

    arXiv 2023

  4. [4]

    and Guo, B

    Babu s ka, I. and Guo, B. (1992). ``The h, p and hp version of the finite element method; basis theory and applications.''\ Advances in Engineering Software , 15(3-4), 159--174

    work page 1992

  5. [5]

    A., and Katz, I

    Babuska, I., Szabo, B. A., and Katz, I. N. (1981). ``The p-version of the finite element method.''\ SIAM journal on numerical analysis , 18(3), 515--545

    work page 1981

  6. [6]

    Brunton, S. L. and Kutz, J. N. (2022). Data-driven science and engineering: Machine learning, dynamical systems, and control . Cambridge University Press

    2022

  7. [7]

    Cai, S., Mao, Z., Wang, Z., Yin, M., and Karniadakis, G. E. (2022). ``Physics-informed neural networks (pinns) for fluid mechanics: A review.''\ Acta Mechanica Sinica , 1--12

    work page 2022

  8. [8]

    Duarte, C. A. and Oden, J. T. (1996). ``An hp adaptive method using clouds.''\ Computer methods in applied mechanics and engineering , 139(1-4), 237--262

    work page 1996

Show all 38 references
  1. [9]

    Farrar, C. R. and Worden, K. (2012). Structural health monitoring: a machine learning perspective . John Wiley & Sons

    2012

  2. [10]

    Freno, B. A. and Carlberg, K. T. (2019). ``Machine-learning error models for approximate solutions to parameterized systems of nonlinear equations.''\ Computer Methods in Applied Mechanics and Engineering , 348, 250--296

    2019

  3. [11]

    and Ghahramani, Z

    Gal, Y. and Ghahramani, Z. (2016). ``Dropout as a bayesian approximation: Representing model uncertainty in deep learning.''\ international conference on machine learning , PMLR, 1050--1059

    2016

  4. [12]

    Gudivada, V., Apon, A., and Ding, J. (2017). ``Data quality considerations for big data and machine learning: Going beyond data cleaning and transformations.''\ International Journal on Advances in Software , 10(1), 1--20

    2017

  5. [13]

    and Tarvainen, T

    Hauptmann, A. and Tarvainen, T. (2023). ``Model-based reconstructions for quantitative imaging in photoacoustic tomography.''\ arXiv preprint arXiv:2311.15735

    2023 arXiv

  6. [14]

    and Ding, Q

    He, M.-X. and Ding, Q. (2021). ``Data-driven optimization of the periodic beam with multiple acoustic black holes.''\ Journal of Sound and Vibration , 493, 115816

    2021

  7. [15]

    Q., Vu, Q.-V., and Nguyen-Xuan, H

    Hoang, V.-N., Nguyen, N.-L., Tran, D. Q., Vu, Q.-V., and Nguyen-Xuan, H. (2022). ``Data-driven geometry-based topology optimization.''\ Structural and Multidisciplinary Optimization , 65(2), 1--15

    2022

  8. [16]

    and Somersalo, E

    Kaipio, J. and Somersalo, E. (2006). Statistical and computational inverse problems , Vol. 160. Springer Science & Business Media

    2006

  9. [17]

    E., Kevrekidis, I

    Karniadakis, G. E., Kevrekidis, I. G., Lu, L., Perdikaris, P., Wang, S., and Yang, L. (2021). ``Physics-informed machine learning.''\ Nature Reviews Physics , 3(6), 422--440

    2021

  10. [18]

    Koponen, J., L \"a hivaara, T., Kaipio, J., and Vauhkonen, M. (2021). ``Model reduction in acoustic inversion by artificial neural network.''\ The Journal of the Acoustical Society of America , 150(5), 3435--3444

    2021

  11. [19]

    Krysl, P. (2010). ``Thermal and stress analysis with the finite element method.''\ Accompanied by the MATLABr toolbox FAESOR. Pressure Cooker Press: San Diego

    2010

  12. [20]

    Lee, S., Ha, J., Zokhirova, M., Moon, H., and Lee, J. (2018). ``Background information of deep learning for structural engineering.''\ Archives of Computational Methods in Engineering , 25(1), 121--129

    2018

  13. [21]

    Lei, X., Liu, C., Du, Z., Zhang, W., and Guo, X. (2019). ``Machine learning-driven real-time topology optimization under moving morphable component-based framework.''\ Journal of Applied Mechanics , 86(1), 011004

    2019

  14. [22]

    Liu, Z., Wu, C., and Koishi, M. (2019). ``A deep material network for multiscale topology learning and accelerated nonlinear modeling of heterogeneous materials.''\ Computer Methods in Applied Mechanics and Engineering , 345, 1138--1168

    2019

  15. [23]

    Liu, Z., Xia, L., Xia, Q., and Shi, T. (2020). ``Data-driven design approach to hierarchical hybrid structures with multiple lattice configurations.''\ Structural and Multidisciplinary Optimization , 61(6), 2227--2235

    2020

  16. [24]

    and Molinaro, R

    Mishra, S. and Molinaro, R. (2021). ``Estimates on the generalization error of physics-informed neural networks for approximating a class of inverse problems for PDEs .''\ IMA Journal of Numerical Analysis

    2021

  17. [25]

    Nissinen, A., Heikkinen, L., and Kaipio, J. (2007). ``The bayesian approximation error approach for electrical impedance tomography—experimental results.''\ Measurement Science and Technology , 19(1), 015501

    2007

  18. [26]

    Raissi, M., Perdikaris, P., and Karniadakis, G. E. (2019). ``Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations.''\ Journal of Computational Physics , 378, 686--707

    2019

  19. [27]

    Reddy, J. N. (2019). Introduction to the finite element method . McGraw-Hill Education

    2019

  20. [28]

    and Valtorta, M

    Sessions, V. and Valtorta, M. (2006). ``The effects of data quality on machine learning algorithms..''\ ICIQ , 6, 485--498

    2006

  21. [29]

    Shi, S. (2020). ``Data-driven structural design optimization for petal-shaped auxetics using isogeometric analysis.''\ Computer Modeling in Engineering & Sciences , 122(2), 433--458

    2020

  22. [30]

    and Kim, Y

    Shin, D. and Kim, Y. Y. (2020). ``Data-driven approach for a one-dimensional thin-walled beam analysis.''\ Computers & Structures , 231, 106207

    2020

  23. [31]

    N., Black, J

    Smyl, D., Tallman, T. N., Black, J. A., Hauptmann, A., and Liu, D. (2021). ``Learning and correcting non-gaussian model errors.''\ Journal of Computational Physics , 432, 110152

    2021

  24. [32]

    Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., and Salakhutdinov, R. (2014). ``Dropout: a simple way to prevent neural networks from overfitting.''\ The Journal of Machine Learning Research , 15(1), 1929--1958

    2014

  25. [33]

    V., and Huang, H

    Sun, H., Burton, H. V., and Huang, H. (2020). ``Machine learning applications for building structural design and performance assessment: state-of-the-art review.''\ Journal of Building Engineering , 101816

    2020

  26. [34]

    Surana, K. S. and Reddy, J. N. (2016). The finite element method for boundary value problems: mathematics and computations . CRC press

    2016

  27. [35]

    Thai, H.-T. (2022). ``Machine learning for structural engineering: A state-of-the-art review.''\ Structures , Vol. 38, Elsevier, 448--491

    2022

  28. [36]

    Xue, L., Liu, J., Wen, G., and Wang, H. (2021). ``Efficient, high-resolution topology optimization method based on convolutional neural networks.''\ Frontiers of Mechanical Engineering , 16(1), 80--96

    2021

  29. [37]

    Zhou, J., Cui, G., Hu, S., Zhang, Z., Yang, C., Liu, Z., Wang, L., Li, C., and Sun, M. (2020). ``Graph neural networks: A review of methods and applications.''\ AI open , 1, 57--81

    2020

  30. [38]

    A., Ramaswamy, H., Meyer, R., Sinkov, A., and Good, M

    Zhuang, B., Gencturk, B., Oberai, A. A., Ramaswamy, H., Meyer, R., Sinkov, A., and Good, M. (2024). ``Impurity gas detection for snf canisters using probabilistic deep learning and acoustic sensing.''\ Measurement Science and Technology , 35(12), 126005

    2024

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.