Generative AI-enhanced Probabilistic Multi-Fidelity Surrogate Modeling Via Transfer Learning

David Barajas-Solano; Hui Chen; Jice Zeng

REVIEW 3 major objections 5 minor 30 references

Generative AI-enhanced Probabilistic Multi-Fidelity Surrogate Modeling Via Transfer Learning

T0 review · 3 major / 5 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read A two-stage normalizing-flow surrogate aims to match high-fidelity accuracy using mostly cheap low-fidelity simulations.

desk verdict A coherent application of SurVAE/Funnels to multi-fidelity surrogate modeling, but the empirical case is thin and the uncertainty claims are unvalidated. read the letter →

arxiv 2602.00072 v2 pith:CIEXYVZ2 submitted 2026-01-20 cs.LG stat.ML

Jice Zeng , David Barajas-Solano , Hui Chen This is my paper · ORCID

classification cs.LGstat.ML

keywords multi-fidelitymodelingsurrogatenormalizingflowstransferlearninguncertaintyquantificationsurjectivelayersstructuraldynamicsgenerativemodels

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 tries to solve a practical data-scarcity problem: high-fidelity simulations are too expensive to generate in the quantities deep-learning surrogates need, while low-fidelity simulations are cheap but biased. Its proposal is a two-stage training scheme: first pretrain a normalizing-flow generative model on abundant low-fidelity data, then fine-tune it on a small high-fidelity dataset so the flow learns to correct the low-fidelity bias. The central mechanism is a surjective (dimension-reducing) flow layer that compresses high-dimensional time-series responses into a 10-dimensional latent code while preserving exact-likelihood training. If the proposal works, engineers could substitute mostly cheap coarse-mesh simulations plus a small fine-mesh set for a large fine-mesh dataset, and still obtain probabilistic predictions with quantified uncertainty rather than point estimates. The paper demonstrates this on a reinforced-concrete slab benchmark under both strong and weak low-fidelity/high-fidelity correlation, reporting high-fidelity-level accuracy in both regimes.

What carries the argument

Central object: a surjective normalizing flow, a neural density model that maps data to a simple base distribution through invertible layers with exact likelihoods. Since standard flows preserve dimension, the paper inserts a dimension-reducing 'funnel' layer that splits each response into kept and discarded blocks, maps the kept block to a 10-dimensional latent code with a coupling transform, and models the discarded block with a Gaussian decoder. The exact log-likelihood becomes base log-density plus decoder log-density plus a Jacobian correction. This compression makes large-scale low-fidelity pretraining and small high-fidelity fine-tuning tractable, and fine-tuning transfers knowledge b

What would settle it

Concrete check: hold out high-fidelity responses from a time segment where the signal is sharply transient, fine-tune the flow on the remaining responses, and count how often the claimed 95% credibility interval actually contains the simulated high-fidelity response; if coverage falls far below 95% in those segments, or if varying the latent dimension from small to large never reproduces the reported accuracy, the central claim is called into question.

Watch

Extended reading notes

Core claim

The paper claims that a surjective normalizing flow, pretrained on abundant low-fidelity data and then fine-tuned on a small high-fidelity dataset, produces probabilistic predictions with high-fidelity accuracy while using far fewer high-fidelity simulations. The key architectural idea is a dimension-reducing 'funnel' layer that compresses 200-step response time series to a 10-dimensional latent code while keeping the exact log-likelihood computable. In the reinforced-concrete slab benchmark, the fine-tuned model tracks the high-fidelity response closely with narrow credibility intervals, even under weak low-fidelity/high-fidelity correlation, and the paper reports that it beats a determinis

Load-bearing premise

The approach stands on the assumption that a 10-dimensional latent code, plus a Gaussian decoder for the discarded part of the response, is expressive enough to represent the difference between low- and high-fidelity behavior for unseen inputs; the paper fixes this capacity without a sensitivity study.

Editorial extensions

If this is right

If the central claim holds, expensive fine-mesh simulations can be largely replaced by coarse-mesh simulations plus a modest fine-mesh set, cutting the cost of building probabilistic surrogates for structural dynamics.
The same pretrain-then-fine-tune recipe should apply to other high-dimensional responses, such as fields or long time series, wherever standard bijective flows are impractical due to their dimension-preserving constraint.
Downstream tasks that need uncertainty—reliability analysis, Bayesian updating, risk-informed decisions—can use full predictive distributions from the surrogate instead of point estimates.
Because the paper's ablation shows that pretraining on low-fidelity data helps even when high-fidelity data is scarce, the approach offers a concrete answer to the data-hunger of deep surrogates.

Reading between the lines

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

A natural next test the paper does not run: measuring calibration of the 95% credibility intervals (does the true high-fidelity response fall inside them about 95% of the time?), which would tell whether the uncertainty estimates are trustworthy, not just narrow.
The same two-stage scheme could likely be coupled with active learning—choosing which high-fidelity simulations to run next—to reduce the required fine-tuning set further; the paper does not explore this.
The 10-dimensional latent code may itself encode physical structure (e.g., dominant modes of the response); inspecting it could give interpretability and a diagnostic for when transfer is working.
One could test the framework when low- and high-fidelity models differ by simplified physics rather than mesh refinement, a harder and arguably more realistic transfer scenario than the excitation perturbation used in the paper's Case 2.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Colleague,

The paper is a straightforward combination of published pieces: surjective normalizing flows (SurVAE/Funnels) as the conditional density model, and transfer learning from low-fidelity to high-fidelity data. The math in Sec. 3 is internally consistent; the likelihood decomposition is a restatement of Nielsen et al. and Klein et al. The new part is the application: using this to build multi-fidelity surrogates for a reinforced concrete slab benchmark, with different LF-HF correlation regimes. That is a legitimate new application, and the framework is plausible.

What the paper does well: it is clearly written, the model description is precise, and the authors are honest that the components are not new. The empirical demonstration, while limited, does show the expected behavior: fine-tuned model beats LF-only and HF-only baselines on the few test cases shown. The relative L2 comparison against TCN on 20 test sets is a step in the right direction.

Where it falls short: the evidence is far too thin for the claims. The strongest quantitative results rest on two selected test pairs per case. There is no repeated training, no sensitivity analysis for the latent dimension Q or architecture, and no error bars around the aggregate L2 numbers. More importantly, the paper's central value proposition is probabilistic prediction, but predictive intervals are never validated. No calibration or coverage check appears anywhere; the 95% CIs are shown for two examples and interpreted visually. For a method whose selling point is uncertainty quantification, that is a load-bearing gap. The decoder in Eq. (16) is a diagonal Gaussian over the discarded 190-dimensional block, so conditional independence is assumed; without calibration evidence, the claim of 'reliable uncertainty estimates' is unsupported.

There is also a mismatch: the abstract promises a rail-sleeper-ballast benchmark, but the paper only presents the RC slab. That needs reconciliation.

The math is not the problem. The evaluation is. If the authors release code/data, run more repetitions, report aggregate statistics with variability, add a sensitivity study for Q and architecture, and actually check whether the claimed intervals have the right coverage, this could be a serviceable application paper. As it stands, it is a promising but unproven demonstration.

Referee Report

3 major / 5 minor

Summary. The paper proposes a two-stage probabilistic multi-fidelity surrogate modeling framework. A conditional surjective normalizing flow is first pretrained on abundant low-fidelity (LF) simulation data and then fine-tuned on a small high-fidelity (HF) dataset. To handle high-dimensional time-series responses, the architecture uses a funnel layer that reduces the state from 200 to 10 dimensions while retaining exact log-likelihood training; a diagonal Gaussian decoder models the discarded block. The approach is validated on a reinforced-concrete slab finite-element model under two scenarios (strong and weak LF-HF correlation), including a comparison against a multi-fidelity temporal convolutional network (TCN) baseline. The paper claims HF-accurate probabilistic predictions with reliable uncertainty estimates and a significant improvement over LF-only baselines while using fewer HF evaluations.

Significance. The methodological idea is timely and practically relevant: if validated, it would allow engineers to replace many expensive HF simulations with cheap LF simulations plus a small HF fine-tuning set, while obtaining calibrated probabilistic predictions. The derivation of the surjective-layer likelihood in Sec. 3.2 is explicit and internally consistent, and the benchmark is a nontrivial structural dynamics application. The inclusion of a deterministic TCN baseline is a useful sanity check. However, the empirical evidence presented is not sufficient to establish the central probabilistic claims: no calibration/coverage validation is reported, and the key accuracy comparisons are single-point estimates without statistical spread. The paper would be significantly stronger with repeated runs, uncertainty calibration checks, and sensitivity studies of the core architectural choices.

major comments (3)

[Secs. 4.1–4.2, 5] The paper's central value proposition is probabilistic prediction, yet no calibration or coverage check is reported. The 95% CIs are shown only for two test pairs per case (Figs. 5, 6, 9, 10), and the number of Monte Carlo samples used to compute means and CIs is not stated. Since the decoder in Eq. (16) is a diagonal Gaussian on a 190-dimensional discarded block, the intervals could be systematically miscalibrated even if predictive means are accurate. Please add empirical coverage on the 20 held-out pairs (e.g., fraction of true responses inside the 95% CI, PIT histograms) and report the MC sample count. Without this, the conclusion of 'reliable uncertainty estimates' is unsupported.
[Sec. 4.4, 4.3] The headline comparison rests on a single point estimate: average relative L2 error 0.091 for TCN vs 0.0791 for the proposed method over 20 test datasets. No variance, confidence interval, paired significance test, or repeated runs with different seeds are reported, and TCN hyperparameters are fixed without tuning. Moreover, the abstract claims 'significantly outperforms LF-only baselines,' but no systematic LF-only baseline is included in the error comparisons; only two examples show LF R^2 values. Please add distributions over repeated runs and a statistical test, and clarify what 'fewer HF evaluations' is compared with.
[Sec. 4.1, Eq. (16)] The architecture fixes the latent dimension Q=10 and layer dimensionalities [200,200,200,200,10,10,10] without any sensitivity study. The capacity of the surjective layer and the Gaussian decoder to represent the LF-HF discrepancy depends critically on Q and decoder size. Please include ablations varying Q (e.g., 5, 20, 50) or at least a discussion of how Q was selected. As it stands, the reported HF accuracy may be configuration-specific rather than a general property of the framework.

minor comments (5)

[Abstract] There are two abstracts in the manuscript. The first claims validation on 'a rail-sleeper-ballast and a reinforced concrete slab,' while the full-text abstract and the body present only the reinforced-concrete slab benchmark. This inconsistency must be resolved.
[Sec. 3.2, Eq. (8)] Equation (8) is written as a pointwise equality, but the right-hand side depends on a sample z ~ q(z|y,theta). In the SurVAE framework this is an expectation over q or a lower-bound expression. Please clarify the notation so that the subsequent derivation is unambiguous.
[Sec. 3.1] The heading 'Bijecive conditional NFs' contains a typo; it should be 'Bijective'. Also, Eq. (21) contains a duplicated word ('provides provides').
[Sec. 4.4] The reported 12.7% reduction appears inconsistent with the stated values 0.091 vs 0.0791, which corresponds to approximately 13.1%. Please verify the arithmetic.
[Sec. 4.3] The HF-only model is described as 'training a single NF model using only 180 HF data pairs,' but its architecture, training epochs, and batch size are not specified. State whether it uses the same surjective architecture and training schedule as the proposed method for a fair comparison.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; held-out evaluation and externally grounded likelihood derivation.

full rationale

The paper's derivation chain is self-contained against held-out data. The claimed pipeline (pretrain a surjective normalizing flow on 1,000 LF pairs via Eq. (1), fine-tune on 180 HF pairs via Eq. (2), then evaluate relative L2 error and 95% CIs on the 20 held-out HF test pairs) never defines a predicted quantity in terms of a fitted parameter; the reported test metrics come from unseen data. The surjective-flow likelihood expression, Eq. (15), is attributed to external references [23,25] and is implemented via the external surjectors library [27], not via a same-author chain. The TCN baseline [15,16] is re-trained in the paper on the same data, so the comparison is not imported as a result. Self-citations such as [3,4,6,16] appear, but none is load-bearing for the central claim: no uniqueness theorem from the authors forbids alternatives, and no ansatz is smuggled in by self-citation. The absence of a calibration or coverage check for the claimed 95% credibility intervals is a validation/evidence gap, but it is not circularity—no equation reduces to its own input and no fitted parameter is renamed as a prediction.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central contribution is an empirical framework. The only 'derivation' is the likelihood of a published surrogate architecture; all quantitative claims rest on hand-chosen architecture/hyperparameters and simulated data. No new entities are postulated.

free parameters (6)

Latent dimension Q and layer dimensionalities = Q=10; [200,200,200,200,10,10,10]
The surjective layer reduces 200-dim output to 10-dim latent; the paper asserts this is sufficient compression but provides no sensitivity analysis or criterion for choosing Q.
Funnel layer placement = fifth layer; 95% reduction
Architectural choice (after four bijective layers, before two bijective layers) is motivated only by a heuristic about extracting structure before compression; no ablation.
Case 2 excitation perturbation F_delta = U(-0.6, 0.6)
This perturbation defines the 'weak correlation' scenario; its magnitude is chosen by hand and directly controls how hard the transfer problem is.
HF fine-tuning dataset sizes = 100 and 180 samples (20 held out)
The claim 'fewer HF evaluations' is demonstrated only for 100 vs 180; no learning-curve or cost-benefit analysis.
TCN baseline hyperparameters = 10 blocks, 60 channels, kernel 20, dropout 0.2
No hyperparameter tuning or multiple runs are reported for the baseline, so the 12.7% improvement could reflect baseline underfitting.
Training hyperparameters = Adam lr=1e-4; batch 64/16; epochs 1000/500
Standard choices, but no sensitivity/ablation; the reported numbers depend on them.

assumptions (5)

standard math Surjective layer log-likelihood decomposition (SurVAE/Funnels) is exact under the funnel right-inverse construction.
Eqs. (8)-(15) are reproduced from Nielsen et al. [23] and Klein et al. [25]; the paper relies on their proof of vanishing bound looseness rather than deriving it.
domain assumption HF model's expected prediction error wrt true model G is negligible; LF model is significantly biased; LF sampling is much cheaper.
Assumed in Sec. 2.1; not verified empirically.
domain assumption The conditional normalizing flow can represent the relevant LF and HF conditional distributions at the chosen capacity.
No universal-approximation or convergence statement; the method's success depends on this.
domain assumption Fine-tuning from the LF-trained initialization converges to a good HF solution; gradient descent does not destroy LF knowledge.
Assumed by the transfer-learning strategy in Sec. 2.2; no analysis of the optimization landscape.
domain assumption The FE models (ANSYS, Newmark, Rayleigh damping, band-limited white noise excitation) produce reliable ground-truth data.
Used to generate D_LF and D_HF in Sec. 4; the paper provides no experimental validation against physical measurements.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Generative AI-enhanced Probabilistic Multi-Fidelity Surrogate Modeling Via Transfer Learning." pith.science (2026). https://pith.science/paper/CIEXYVZ2

@misc{pith2026260200072,
  author       = {Pith},
  title        = {Pith review of: Generative AI-enhanced Probabilistic Multi-Fidelity Surrogate Modeling Via Transfer Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CIEXYVZ2}},
  note         = {Machine review of arXiv:2602.00072}
}

read the original abstract

The performance of machine learning surrogates is critically dependent on data quality and quantity. This presents a major challenge, as high-fidelity (HF) data is often scarce and computationally expensive to acquire, while low-fidelity (LF) data is abundant but less accurate. To address this data-scarcity problem, we propose a probabilistic multi-fidelity surrogate modeling framework that integrates transfer learning with generative modeling. We employ a normalizing flow (NF) generative model as the backbone, which is trained in two phases: (i) the NF is first pretrained on a large LF dataset to learn a probabilistic forward model; (ii) the pretrained model is then fine-tuned on a small HF dataset, allowing it to correct for LF--HF discrepancies via knowledge transfer. To relax the dimension-preserving constraint of standard bijective NFs, we integrate surjective (dimension-reducing) layers with standard coupling blocks. This architecture enables learned dimension reduction while preserving the ability to train with exact likelihoods. The resulting surrogate provides fast probabilistic predictions with quantified uncertainty and significantly outperforms LF-only baselines while using fewer HF evaluations. We validate the approach for two benchmark systems: a rail-sleeper-ballast and a reinforced concrete slab. For both applications, we combine many coarse-mesh (LF) simulations with a limited set of fine-mesh (HF) simulations. The proposed model achieves probabilistic predictions with HF accuracy, demonstrating a practical path toward data-efficient, generative AI-driven surrogates for complex engineering systems.

Figures

Figures reproduced from arXiv: 2602.00072 by the authors.

**Figure 2.** FE model of a reinforced-concrete slab: (a) LF model; (b) HF model. [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗

**Figure 3.** Example of simulated LF and HF acceleration responses for Case 1. [PITH_FULL_IMAGE:figures/full_fig_p015_3.png] view at source ↗

Figures from the paper (11 more)

**Figure 4.** Figure 4: Validation of the LF model for Case 1. Next, we proceed to train a MF model building upon the previously trained LF model. The LF model is used as a pre-trained initialization and subsequently fine-tuned using 180 available HF data pairs, while the remaining 20 HF data…

**Figure 5.** Figure 5: MF prediction performance on test data #1 for Case 1: (a) comparison between [PITH_FULL_IMAGE:figures/full_fig_p017_5.png]

**Figure 6.** Figure 6: MF prediction performance on test data #2 for Case 1: (a) comparison between [PITH_FULL_IMAGE:figures/full_fig_p017_6.png]

**Figure 7.** Figure 7: Example of simulated LF and HF acceleration responses for Case 2. [PITH_FULL_IMAGE:figures/full_fig_p018_7.png]

**Figure 8.** Figure 8: presents the validation performance of the trained LF model. On the left, the estimated LF response is compared against the true LF response. The two curves align closely across the full time window, yielding a high R2 = 0.9942. The uncertainty band remains narrow, in…

**Figure 9.** Figure 9: MF prediction performance on test data #1 for Case 2: (a) comparison between [PITH_FULL_IMAGE:figures/full_fig_p020_9.png]

**Figure 10.** Figure 10: MF prediction performance on test data #2 for Case 2: (a) comparison between [PITH_FULL_IMAGE:figures/full_fig_p020_10.png]

**Figure 11.** Figure 11: Effect of different amounts of HF data in Case 1 (strong correlation): (a) Test [PITH_FULL_IMAGE:figures/full_fig_p021_11.png]

**Figure 12.** Figure 12: Effect of different amounts of HF data in Case 2 (weak correlation): (a) Test [PITH_FULL_IMAGE:figures/full_fig_p022_12.png]

**Figure 13.** Figure 13: Comparison between the proposed method (predictive mean and 95% CI) and [PITH_FULL_IMAGE:figures/full_fig_p024_13.png]

**Figure 14.** Figure 14: Probability distributions of predicted HF responses at selected time step for [PITH_FULL_IMAGE:figures/full_fig_p025_14.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 2 linked inside Pith

[1]

H.-R. Bae, R. V. Grandhi, R. A. Canfield, Epistemic uncertainty quan- tification techniques including evidence theory for large-scale structures, Computers & Structures 82 (13-14) (2004) 1101–1112. 26

2004
[2]

Zhao, Y.-M

W. Zhao, Y.-M. Zhang, Reliability analysis of random vibration trans- mission path systems, Mechanical Systems and Signal Processing 113 (2018) 77–89

2018
[3]

J. Zeng, Y. Wang, A. M. Tartakovsky, D. A. Barajas-Solano, Solving high-dimensional inverse problems using amortized likelihood-free infer- ence with noisy and incomplete data, Computer Methods in Applied Mechanics and Engineering 443 (2025) 118064

2025
[4]

J. Zeng, M. D. Todd, Z. Hu, Probabilistic damage detection using a new likelihood-free bayesian inference method, Journal of Civil Structural Health Monitoring 13 (2) (2023) 319–341

2023
[5]

Y. Zong, D. Barajas-Solano, A. M. Tartakovsky, Randomized physics- informed neural networks for bayesian data assimilation, Computer Methods in Applied Mechanics and Engineering 436 (2025) 117670

2025
[6]

J. Zeng, Y. H. Kim, S. Qin, Bayesian model updating for structural dynamic applications combing differential evolution adaptive metropolis and kriging model, Journal of Structural Engineering 149 (6) (2023) 04023070

2023
[7]

Wan, W.-X

H.-P. Wan, W.-X. Ren, Stochastic model updating utilizing bayesian approach and gaussian process model, Mechanical Systems and Signal Processing 70 (2016) 245–268

2016
[8]

Kudela, R

J. Kudela, R. Matousek, Recent advances and applications of surrogate models for finite element method computations: a review: J. kudela, r. matousek, Soft Computing 26 (24) (2022) 13709–13733

2022

Show all 30 references

[9]

Samadian, I

D. Samadian, I. B. Muhit, N. Dawood, Application of data-driven sur- rogate models in structural engineering: a literature review, Archives of Computational Methods in Engineering 32 (2) (2025) 735–784

2025
[10]

Motamed, A multi-fidelity neural network surrogate sampling method for uncertainty quantification, International Journal for Uncer- tainty Quantification 10 (4) (2020)

M. Motamed, A multi-fidelity neural network surrogate sampling method for uncertainty quantification, International Journal for Uncer- tainty Quantification 10 (4) (2020)

2020
[11]

Peherstorfer, K

B. Peherstorfer, K. Willcox, M. Gunzburger, Survey of multifidelity methods in uncertainty propagation, inference, and optimization, Siam Review 60 (3) (2018) 550–591. 27

2018
[12]

Feng, S.-Z

D.-C. Feng, S.-Z. Chen, E. Taciroglu, Deep learning-enhanced efficient seismic analysis of structures with multi-fidelity modeling strategies, Computer Methods in Applied Mechanics and Engineering 421 (2024) 116775

2024
[13]

Stavropoulou, E

C. Stavropoulou, E. Katsidoniotaki, N. Faedo, M. Göteman, Multi- fidelity surrogate modeling of nonlinear dynamic responses in wave en- ergy farms, Applied Energy 380 (2025) 125011

2025
[14]

S. Yang, Y. Lee, N. Kang, Data-efficient deep operator network for un- steady flow: A multi-fidelity approach with physics-guided subsampling, Computer Methods in Applied Mechanics and Engineering 446 (2025) 118254

2025
[15]

Zhong, D.-C

Q.-M. Zhong, D.-C. Feng, S.-Z. Chen, Multi-fidelity enhanced few-shot time series prediction model for structural dynamics analysis, Computer Methods in Applied Mechanics and Engineering 434 (2025) 117583

2025
[16]

J. Zeng, G. Li, Z. Gao, Y. Li, S. Sundararajan, S. Barbat, Z. Hu, Ma- chine learning enabled fusion of cae data and test data for vehicle crash- worthiness performance evaluation by analysis, Structural and Multidis- ciplinary Optimization 66 (4) (2023) 96

2023
[17]

G.-F. He, P. Zhang, Z.-Y. Yin, Active learning inspired multi-fidelity probabilistic modelling of geomaterial property, Computer Methods in Applied Mechanics and Engineering 432 (2024) 117373

2024
[18]

Wang, S.-P

L. Wang, S.-P. Zhu, B. Wu, Z. Xu, C. Luo, Q. Wang, Multi-fidelity physics-informed machine learning framework for fatigue life prediction of additive manufactured materials, Computer Methods in Applied Me- chanics and Engineering 439 (2025) 117924

2025
[19]

F. Xie, X. Zhang, S. Wu, T. Ji, Y. Zheng, Multi-fidelity bayesian neural networks for aerodynamic data fusion with heterogeneous uncertainties, Computer Methods in Applied Mechanics and Engineering 435 (2025) 117666

2025
[20]

Z. Wang, Q. She, T. E. Ward, Generative adversarial networks in computer vision: A survey and taxonomy, ACM Computing Surveys (CSUR) 54 (2) (2021) 1–38. 28

2021
[21]

Iorliam, J

A. Iorliam, J. A. Ingio, A comparative analysis of generative artificial intelligence tools for natural language processing, Journal of Computing Theories and Applications 1 (3) (2024) 311–325

2024
[22]

G.Papamakarios, E.Nalisnick, D.J.Rezende, S.Mohamed, B.Lakshmi- narayanan, Normalizing flows for probabilistic modeling and inference, Journal of Machine Learning Research 22 (57) (2021) 1–64

2021
[23]

Nielsen, P

D. Nielsen, P. Jaini, E. Hoogeboom, O. Winther, M. Welling, Survae flows: Surjections to bridge the gap between vaes and flows, Advances in Neural Information Processing Systems 33 (2020) 12685–12696

2020
[24]

Dirmeier, C

S. Dirmeier, C. Albert, F. Perez-Cruz, Simulation-based inference for high-dimensional data using surjective sequential neural likelihood esti- mation, in: The41stConferenceonUncertaintyinArtificialIntelligence, 2023, p. none

2023
[25]

Klein, J

S. Klein, J. A. Raine, S. Pina-Otey, S. Voloshynovskiy, T. Golling, Fun- nels: Exact maximum likelihood with dimensionality reduction, arXiv preprint arXiv:2112.08069 (2021)

2021 arXiv
[26]

L. Dinh, J. Sohl-Dickstein, S. Bengio, Density estimation using real nvp, arXiv preprint arXiv:1605.08803 (2016)

2016 arXiv
[27]

Dirmeier, Surjectors: surjection layers for density estimation with normalizing flows, Journal of Open Source Software 9 (94) (2024) 6188

S. Dirmeier, Surjectors: surjection layers for density estimation with normalizing flows, Journal of Open Source Software 9 (94) (2024) 6188

2024
[28]

I. K. M. Jais, A. R. Ismail, Adam optimization algorithm for wide and deep neural network, Knowledge Engineering and Data Science 2 (1) (2019) 10

2019
[29]

Hewage, A

P. Hewage, A. Behera, M. Trovati, E. Pereira, M. Ghahremani, F. Palmieri, Y. Liu, Temporal convolutional neural (tcn) network for an effective weather forecasting using time-series data from the local weather station, Soft Computing 24 (21) (2020) 16453–16482

2020
[30]

J. Fan, K. Zhang, Y. Huang, Y. Zhu, B. Chen, Parallel spatio-temporal attention-based tcn for multivariate time series prediction, Neural Com- puting and Applications 35 (18) (2023) 13109–13118. 29

2023

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.

[1] [1]

H.-R. Bae, R. V. Grandhi, R. A. Canfield, Epistemic uncertainty quan- tification techniques including evidence theory for large-scale structures, Computers & Structures 82 (13-14) (2004) 1101–1112. 26

2004

[2] [2]

Zhao, Y.-M

W. Zhao, Y.-M. Zhang, Reliability analysis of random vibration trans- mission path systems, Mechanical Systems and Signal Processing 113 (2018) 77–89

2018

[3] [3]

J. Zeng, Y. Wang, A. M. Tartakovsky, D. A. Barajas-Solano, Solving high-dimensional inverse problems using amortized likelihood-free infer- ence with noisy and incomplete data, Computer Methods in Applied Mechanics and Engineering 443 (2025) 118064

2025

[4] [4]

J. Zeng, M. D. Todd, Z. Hu, Probabilistic damage detection using a new likelihood-free bayesian inference method, Journal of Civil Structural Health Monitoring 13 (2) (2023) 319–341

2023

[5] [5]

Y. Zong, D. Barajas-Solano, A. M. Tartakovsky, Randomized physics- informed neural networks for bayesian data assimilation, Computer Methods in Applied Mechanics and Engineering 436 (2025) 117670

2025

[6] [6]

J. Zeng, Y. H. Kim, S. Qin, Bayesian model updating for structural dynamic applications combing differential evolution adaptive metropolis and kriging model, Journal of Structural Engineering 149 (6) (2023) 04023070

2023

[7] [7]

Wan, W.-X

H.-P. Wan, W.-X. Ren, Stochastic model updating utilizing bayesian approach and gaussian process model, Mechanical Systems and Signal Processing 70 (2016) 245–268

2016

[8] [8]

Kudela, R

J. Kudela, R. Matousek, Recent advances and applications of surrogate models for finite element method computations: a review: J. kudela, r. matousek, Soft Computing 26 (24) (2022) 13709–13733

2022