pith. machine review for the scientific record. sign in

arxiv: 2604.17575 · v1 · submitted 2026-04-19 · 💻 cs.CE

Recognition: unknown

μ-FlowNet: A Deep Learning Approach for Mapping Flow Fields in Irregular Microchannels Using an Attention-based U-Net Encoder-Decoder Architecture

Ganesh Sahadeo Meshram , Suman Chakraborty , Nishant Sinha , Partha Pratim Chakrabarti
Authors on Pith no claims yet

Pith reviewed 2026-05-10 04:46 UTC · model grok-4.3

classification 💻 cs.CE
keywords microfluidicsflow field predictionU-Netattention mechanismdeep learningCFD simulationirregular microchannelsfluid dynamics
0
0 comments X

The pith

An attention-based U-Net predicts fluid flow patterns in irregular microchannels from CFD data more accurately than standard variants.

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

The paper introduces μ-FlowNet to map flow fields through random-shaped microchannels using deep learning instead of slow traditional CFD simulations. Datasets are created via extensive CFD runs on various irregular geometries, then pre-processed for spatial features before training. Three U-Net-based models are compared, and the version with an attention mechanism achieves the highest accuracy scores on held-out test cases.

Core claim

The central claim is that the attention-augmented U-Net encoder-decoder outperforms both a standard U-Net and a T-Net variant when trained to reconstruct fluid flow fields from input channel geometries, reaching a Dice score of 0.9317, an IoU of 0.8731, and the best structural similarity index on CFD-generated test data for unseen random circular microchannel shapes.

What carries the argument

The attention mechanism inside the U-Net encoder-decoder, which selectively weights relevant spatial features extracted from pre-processed CFD simulation images to produce accurate flow-field output maps.

If this is right

  • Microfluidics designers can iterate channel shapes and obtain flow predictions orders of magnitude faster than full CFD runs.
  • The same trained model can be applied to additional random geometries not present in the original training set.
  • Attention weighting improves capture of localized flow features such as recirculation zones compared with plain U-Net architectures.
  • The approach replaces expensive per-geometry simulations with a single training stage followed by rapid inference.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same architecture could be retrained on experimental rather than simulated data to create hybrid models that correct for simulation-reality gaps.
  • Extending the input to include time sequences would allow prediction of transient flow evolution inside evolving channel networks.
  • Coupling the fast flow predictor with an optimization loop could automate discovery of channel geometries that achieve target flow properties.

Load-bearing premise

That CFD simulations of random-shaped channels produce training data representative enough of real microchannel flows for the model to generalize to new geometries without overfitting to simulation artifacts.

What would settle it

Fabricate physical irregular microchannels, measure real flow velocities with particle image velocimetry, and test whether the model's predictions match the experimental fields within the reported error bounds.

read the original abstract

In the complex domain of microfluidics systems, analysing fluid flow patterns through random-shaped circular microchannels is significantly challenging task. Conventional approach of solving such problems using computational fluid dynamics often incapable due to their intensive computational requirements and high simulation times. In this study, addressing these limitations, we introduce $\mu$-FlowNet, a deep learning framework based on the adaptable U-Net autoencoders. This model provides a data-driven approach that enhances the prediction and mapping of random-shaped circular microchannels and their corresponding fluid flow patterns. The datasets required for the training of the model is generated by performing extensive simulations using conventional approach of computational fluid dynamics methods. The datasets are then pre-processed and accessed the required spatial and temporal features that are essential for the training. We have trained three different models based on U-Net framework namely, standard U-Net, T-Net, and U-Net with attention mechanism to compare the prediction accuracy and loss. The accuracy of the $\mu$-FlowNet is compared using metrics of dice score and intersection over union and it shows that U-Net with attention mechanism shows the highest dice score and IoU of 0.9317 and 0.8731, respectively and shows the highest structural similarity as compared to standard U-Net and T-Net. This show that U-Net with attention mechanism serves best model to map the fluid flow pattern with random datasets on testing.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper introduces μ-FlowNet, an attention-based U-Net encoder-decoder for predicting fluid flow patterns in random-shaped microchannels. Training data comes from CFD simulations; three U-Net variants (standard, T-Net, attention-augmented) are trained and compared. The central claim is that the attention model is superior, attaining Dice score 0.9317, IoU 0.8731, and the highest SSIM on test data.

Significance. If the performance ranking can be substantiated with regression-appropriate metrics and full experimental details, the work would demonstrate a viable data-driven surrogate for expensive CFD in microfluidics design, enabling faster iteration over irregular geometries. The attention mechanism is a plausible architectural choice for capturing spatial flow features, but the current evidence does not yet establish this advantage.

major comments (2)
  1. [Abstract] Abstract: Dice score and IoU are reported as primary accuracy metrics for 'mapping fluid flow patterns,' yet these quantities are defined for binary segmentation masks. The manuscript provides no description of output encoding (scalar velocity/pressure fields versus masks), loss function, binarization threshold, or post-processing that would convert continuous regression targets into the binary overlap measured by Dice/IoU. Without this, the numerical superiority cannot be interpreted as improved physical flow prediction.
  2. [Methods] Methods/Dataset description (inferred from abstract and results): no information is given on total number of CFD simulations, train/validation/test split ratios, distribution of random channel shapes, hyperparameter search, or any statistical test for the reported metric differences. These omissions leave open the possibility that the 0.9317/0.8731 figures reflect limited validation or post-hoc selection rather than robust generalization.
minor comments (2)
  1. [Title and Abstract] The title refers to 'Mapping Flow Fields' while the evaluation relies on segmentation metrics; a brief clarification of the precise output variable (velocity magnitude, pressure, or derived mask) would improve readability.
  2. [Dataset generation] No mention of the CFD solver, boundary conditions, or Reynolds-number range used to generate the training data; adding these details would help readers assess domain coverage.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed feedback. We address each major comment below and have revised the manuscript to improve clarity on metric application and experimental reproducibility.

read point-by-point responses

  1. Referee: [Abstract] Abstract: Dice score and IoU are reported as primary accuracy metrics for 'mapping fluid flow patterns,' yet these quantities are defined for binary segmentation masks. The manuscript provides no description of output encoding (scalar velocity/pressure fields versus masks), loss function, binarization threshold, or post-processing that would convert continuous regression targets into the binary overlap measured by Dice/IoU. Without this, the numerical superiority cannot be interpreted as improved physical flow prediction.

    Authors: We agree that the manuscript lacks an explicit description of how continuous flow fields are evaluated with segmentation metrics. The flow field mapping is formulated by encoding the CFD-derived velocity data as binary masks that distinguish flow regions from background, enabling direct application of Dice and IoU. We will add a new paragraph in the Methods section detailing the output representation, the loss function used during training, the binarization procedure, and any post-processing. This revision will allow readers to interpret the reported scores as measures of predicted flow pattern fidelity. revision: yes

  2. Referee: [Methods] Methods/Dataset description (inferred from abstract and results): no information is given on total number of CFD simulations, train/validation/test split ratios, distribution of random channel shapes, hyperparameter search, or any statistical test for the reported metric differences. These omissions leave open the possibility that the 0.9317/0.8731 figures reflect limited validation or post-hoc selection rather than robust generalization.

    Authors: We acknowledge that these experimental details are missing from the current text. We will expand the Methods and Dataset sections to report the total number of CFD simulations performed, the train/validation/test split ratios, the procedure and parameter ranges used to generate the random channel shapes, the hyperparameter selection process, and the outcome of statistical significance tests comparing the three models. These additions will substantiate the robustness of the performance ranking. revision: yes

Circularity Check

0 steps flagged

No circularity; empirical training and held-out evaluation on independent CFD data

full rationale

The paper generates datasets via separate CFD simulations, preprocesses them, trains three U-Net variants (standard, T-Net, attention-augmented) in a standard supervised manner, and reports performance on test data using Dice, IoU, and SSIM. No equation, claim, or result reduces by construction to a fitted parameter, self-defined quantity, or self-citation chain. The performance ranking is an empirical observation on held-out data, not a tautology. The metric choice (Dice/IoU on flow fields) may be questionable for regression but does not create circularity in the derivation.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that CFD simulations supply reliable ground-truth labels and on standard supervised learning assumptions that a convolutional network can learn the geometry-to-flow mapping from finite training examples.

free parameters (1)
  • U-Net architecture hyperparameters
    Number of layers, filter counts, and attention module settings chosen during model development to maximize reported metrics.
axioms (1)
  • domain assumption CFD simulations accurately represent the target fluid flow physics for training purposes
    The paper generates all training data via conventional CFD methods without independent experimental validation mentioned.

pith-pipeline@v0.9.0 · 5572 in / 1364 out tokens · 49586 ms · 2026-05-10T04:46:53.566947+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

51 extracted references · 41 canonical work pages

  1. [1]

    Enhanced Fluid Mixing in Microchannels Using Levitated Magnetic Microrobots: A Numerical Study,

    A. A. Demircali, A. Yilmaz, and H. Uvet, “Enhanced Fluid Mixing in Microchannels Using Levitated Magnetic Microrobots: A Numerical Study,” Micromachines, vol. 16, no. 1, Art. no. 1, Jan. 2025, doi: 10.3390/mi16010052

    work page doi:10.3390/mi16010052 2025

  2. [2]

    REVIEW PDMS microfluidics: A mini review,

    R. M. Kiran and S. Chakraborty, “REVIEW PDMS microfluidics: A mini review,” 2020, doi: 10.1002/app.48958

    work page doi:10.1002/app.48958 2020

  3. [3]

    Chakraborty, Microfluidics and Microscale Transport Processes

    S. Chakraborty, Microfluidics and Microscale Transport Processes. CRC Press, 2012

    2012

  4. [4]

    Surface Roughness Analysis of Microchannels Featuring Microfluidic Devices Fabricated by Three Different Materials and Methods,

    J. M. Acosta-Cuevas, M. A. García-Ramírez, G. Hinojosa-Ventura, Á. J. Martínez-Gómez, V. H. Pérez -Luna, and O. González -Reynoso, “Surface Roughness Analysis of Microchannels Featuring Microfluidic Devices Fabricated by Three Different Materials and Methods,” Coatings, vol. 13, no. 10, p. 1676, Sep. 2023, doi: 10.3390/coatings13101676

    work page doi:10.3390/coatings13101676 2023

  5. [5]

    The Effect of Surface Roughness on the Contact Line and Splashing Dynamics of Impacting Droplets,

    M. A. Quetzeri-Santiago, A. A. Castrejón-Pita, and J. R. Castrejón-Pita, “The Effect of Surface Roughness on the Contact Line and Splashing Dynamics of Impacting Droplets,” Sci. Rep., vol. 9, no. 1, p. 15030, Oct. 2019, doi: 10.1038/s41598-019-51490-5

    work page doi:10.1038/s41598-019-51490-5 2019

  6. [6]

    Numerical investigation on pressure-driven electro osmatic flow and mixing in a constricted micro channel by triangular obstacle,

    A. S. C et al., “Numerical investigation on pressure-driven electro osmatic flow and mixing in a constricted micro channel by triangular obstacle,” Int. J. Numer. Methods Heat Amp Fluid Flow, vol. 31, no. 3, pp. 982–1013, Jul. 2020, doi: 10.1108/HFF-06-2020-0349

    work page doi:10.1108/hff-06-2020-0349 2020

  7. [7]

    Numerical analysis of electroosmotic mixing in a heterogeneous charged micromixer with obstacles,

    B. Mondal, S. K. Mehta, S. Pati, and P. K. Patowari, “Numerical analysis of electroosmotic mixing in a heterogeneous charged micromixer with obstacles,” Chem. Eng. Process. - Process Intensif., vol. 168, p. 108585, Nov. 2021, doi: 10.1016/j.cep.2021.108585

    work page doi:10.1016/j.cep.2021.108585 2021

  8. [8]

    Heat transfer and pressure drop in microchannels with random roughness,

    N. Pelević and Th. H. van der Meer, “Heat transfer and pressure drop in microchannels with random roughness,” Int. J. Therm. Sci. , vol. 99, pp. 125 –135, Jan. 2016, doi: 10.1016/j.ijthermalsci.2015.08.012

    work page doi:10.1016/j.ijthermalsci.2015.08.012 2016

  9. [9]

    Water droplet bouncing on a hierarchical superhydrophobic surface fabricated by hydrothermal synthesis and ultraprecision machining,

    Y. Fu, F. Zhao, Z. Zhao, H. Tang, S. To, and C. -H. Cheng, “Water droplet bouncing on a hierarchical superhydrophobic surface fabricated by hydrothermal synthesis and ultraprecision machining,” J. Adhes. Sci. Technol. , vol. 37, no. 4, pp. 706 –720, Feb. 2023, doi: 10.1080/01694243.2022.2038995

    work page doi:10.1080/01694243.2022.2038995 2023

  10. [10]

    Deep learning assisted mechanotyping of individual cells through repeated deformations and relaxations in undulating channels,

    C. Combs, D. D. Seith, M. J. Bovyn, S. P. Gross, X. Xie, and Z. S. Siwy, “Deep learning assisted mechanotyping of individual cells through repeated deformations and relaxations in undulating channels,” Biomicrofluidics, vol. 16, no. 1, 2022

    2022

  11. [11]

    Deep learning models for air quality forecasting based on spatiotemporal characteristics of data,

    K. Rehman, I. Abid, and S. H. Hong, “Deep learning models for air quality forecasting based on spatiotemporal characteristics of data,” Phys. Fluids, vol. 36, no. 5, p. 057111, May 2024, doi: 10.1063/5.0207834

    work page doi:10.1063/5.0207834 2024

  12. [12]

    Mondal, V

    N. Monda l, V. Arya, P. Sarangi, and C. Bakli, “Interplay of roughness and wettability in microchannel fluid flows—Elucidating hydrodynamic details assisted by deep learning,” Phys. Fluids, vol. 36, no. 6, p. 062014, Jun. 2024, doi: 10.1063/5.0208554

    work page doi:10.1063/5.0208554 2024

  13. [13]

    Coupled CFD-DEM Modeling: Formulation, Implementation and Application to Multiphase Flows,

    Hamid Reza Norouzi et al., “Coupled CFD-DEM Modeling: Formulation, Implementation and Application to Multiphase Flows,” Oct. 2016, doi: 10.1002/9781119005315

    work page doi:10.1002/9781119005315 2016

  14. [14]

    Surrogate -Based Aerodynamic Design Optimization: Use of Surrogates in Aerodynamic Design Optimization,

    M. Y. M. Ahmed and N. Qin, “Surrogate -Based Aerodynamic Design Optimization: Use of Surrogates in Aerodynamic Design Optimization,” Int. Conf. Aerosp. Sci. Aviat. Technol. , vol. 13, no. AEROSPACE SCIENCES&AVIATION TECHNOLOGY, ASAT -13, May 26 – 28, 2009, pp. 1–26, May 2009, doi: 10.21608/asat.2009.23442

    work page doi:10.21608/asat.2009.23442 2009

  15. [15]

    Physics Informed Deep Learning (Part I): Data-driven Solutions of Nonlinear Partial Differential Equations

    M. Raissi, P. Perdikaris, and G. E. Karniadakis, “Physics Informed Deep Learning (Part I): Data- driven Solutions of Nonlinear Partial Differential Equations,” Nov. 28, 2017, arXiv: arXiv:1711.10561. doi: 10.48550/arXiv.1711.10561

    work page Pith review doi:10.48550/arxiv.1711.10561 2017

  16. [16]

    Physics -Informed Neural Networks: A Deep Learning Framework for Solving Forward and Inverse Problems Involving Nonlinear Partial Differential Equations,

    M. Raissi, P. Perdikaris, and G. E. Karniadakis, “Physics -Informed Neural Networks: A Deep Learning Framework for Solving Forward and Inverse Problems Involving Nonlinear Partial Differential Equations,” 2018. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0021999118307125

    2018

  17. [17]

    Hidden physics models: Machine learning of nonlinear partial differential equations,

    M. Raissi and G. E . Karniadakis, “Hidden Physics Models: Machine Learning of Nonlinear Partial Differential Equations,” Aug. 2017, doi: 10.1016/j.jcp.2017.11.039

    work page doi:10.1016/j.jcp.2017.11.039 2017

  18. [18]

    Potential of physics -informed neural networks for solving fluid flow problems with parametric boundary conditions,

    F. Lorenzen, A. Zargaran, and U. Janoske, “Potential of physics -informed neural networks for solving fluid flow problems with parametric boundary conditions,” Phys. Fluids, vol. 36, no. 3, p. 037143, Mar. 2024, doi: 10.1063/5.0193952

    work page doi:10.1063/5.0193952 2024

  19. [19]

    Physics -informed generative adversarial networks for stochastic differential equations,

    L. Yang, D. Zhang, and G. E. M. Karniadakis, “Physics -informed generative adversarial networks for stochastic differential equations,” SIAM J. Sci. Comput., vol. 42, no. 1, pp. A292– A317, 2020, doi: 10.1137/18M1225409

    work page doi:10.1137/18m1225409 2020

  20. [20]

    Multistep Neural Networks for Data-driven Discovery of Nonlinear Dynamical Systems

    M. Raissi, P. Perdikaris, and G. E. Karniadakis, “Multistep Neural Networks for Data -driven Discovery of Nonlinear Dynamical Systems,” Jan. 03, 2018, arXiv: arXiv:1801.01236. Accessed: Aug. 13, 2024. [Online]. Available: http://arxiv.org/abs/1801.01236

    work page Pith review arXiv 2018

  21. [21]

    Machine learning of linear differential equations using Gaussian processes,

    M. Raissi, P. Perdikaris, and G. E. Karniadakis, “Machine learning of linear differential equations using Gaussian processes,” J. Comput. Phys., vol. 348, pp. 683–693, Nov. 2017, doi: 10.1016/j.jcp.2017.07.050

    work page doi:10.1016/j.jcp.2017.07.050 2017

  22. [22]

    High -order splitting methods for the incompressible Navier-Stokes equations,

    G. E. Karniadakis, M. Israeli, and S. A. Orszag, “High -order splitting methods for the incompressible Navier-Stokes equations,” J. Comput. Phys., vol. 97, no. 2, pp. 414 –443, Dec. 1991, doi: 10.1016/0021-9991(91)90007-8

    work page doi:10.1016/0021-9991(91)90007-8 1991

  23. [23]

    NSFnets (Navier -Stokes flow nets): Physics - informed neural networks for the incompressible Navier -Stokes equations,

    X. Jin, S. Cai, H. Li, and G. E. Karniadakis, “NSFnets (Navier -Stokes flow nets): Physics - informed neural networks for the incompressible Navier -Stokes equations,” J. Comput. Phys., vol. 426, p. 109951, Feb. 2021, doi: 10.1016/j.jcp.2020.109951

    work page doi:10.1016/j.jcp.2020.109951 2021

  24. [24]

    Quantifying model form uncertainty in Reynolds -averaged turbulence models with Bayesian deep neural networks,

    N. Geneva and N. Zabaras, “Quantifying model form uncertainty in Reynolds -averaged turbulence models with Bayesian deep neural networks,” J. Comput. Phys., vol. 383, pp. 125– 147, 2019, doi: 10.1016/j.jcp.2019.01.021

    work page doi:10.1016/j.jcp.2019.01.021 2019

  25. [25]

    Journal of Fuild Mechanics 807, 155–166 (2016)

    J. Ling, A. Kurzawski, and J. Templeton, “Reynolds averaged turbulence modelling using deep neural networks with embedded invariance,” J. Fluid Mech., vol. 807, pp. 155–166, Nov. 2016, doi: 10.1017/jfm.2016.615

    work page doi:10.1017/jfm.2016.615 2016

  26. [26]

    Prediction of wind shear layer for dynamic soaring by using proper orthogonal decomposition and long short term memory network,

    D. Wang, F. Xie, T. Ji, X. Zhang, Y. Lu, and Y. Zheng, “Prediction of wind shear layer for dynamic soaring by using proper orthogonal decomposition and long short term memory network,” Phys. Fluids, vol. 35, no. 8, 2023

    2023

  27. [27]

    Surfactant-assisted solvothermal synthesis of pure nickel submicron spheres with microwave-absorbing properties,

    H. Guo et al., “Surfactant-assisted solvothermal synthesis of pure nickel submicron spheres with microwave-absorbing properties,” Nanoscale Res. Lett. , vol. 11, no. 1, Dec. 2016, doi: 10.1186/s11671-016-1562-y

    work page doi:10.1186/s11671-016-1562-y 2016

  28. [28]

    Prediction of aerodynamic flow fields using convolutional neural networks,

    S. Bhatnagar, Y. Afshar, S. Pan, K. Duraisamy, and S. Kaushik, “Prediction of aerodynamic flow fields using convolutional neural networks,” Comput. Mech., vol. 64, no. 2, pp. 525 –545, Aug. 2019, doi: 10.1007/s00466-019-01740-0

    work page doi:10.1007/s00466-019-01740-0 2019

  29. [29]

    Prediction of osteoporosis in patients with rheumatoid arthritis using machine learning,

    C. Lee, G. Joo, S. Shin, H. Im, and K. W. Moon, “Prediction of osteoporosis in patients with rheumatoid arthritis using machine learning,” Sci. Rep., vol. 13, no. 1, p. 21800, Dec. 2023, doi: 10.1038/s41598-023-48842-7

    work page doi:10.1038/s41598-023-48842-7 2023

  30. [30]

    doi: 10.2514/1.j058291

    N. Thuerey, K. Weißenow, L. Prantl, and X. Hu, “Deep Learning Methods for Reynolds - Averaged Navier–Stokes Simulations of Airfoil Flows,” AIAA J., vol. 58, no. 1, pp. 25–36, 2020, doi: 10.2514/1.J058291

    work page doi:10.2514/1.j058291 2020

  31. [31]

    Predictions of turbulent shear flows using deep neural networks,

    P. A. Srinivasan, L. Guastoni, H. Azizpour, P. Schlatter, and R. Vinuesa, “Predictions of turbulent shear flows using deep neural networks,” Phys. Rev. Fluids, vol. 4, no. 5, p. 054603, May 2019, doi: 10.1103/PhysRevFluids.4.054603

    work page doi:10.1103/physrevfluids.4.054603 2019

  32. [32]

    Drag prediction of rough- wall turbulent flow using data -driven regression,

    Z. Shi, S. M. H. Khorasani, H. Shin, J. Yang, S. Lee, and S. Bagheri, “Drag prediction of rough- wall turbulent flow using data -driven regression,” May 15, 2024, arXiv: arXiv:2405.09256. Accessed: Sep. 25, 2024. [Online]. Available: http://arxiv.org/abs/2405.09256

    work page arXiv 2024

  33. [33]

    Methods for neural network based prediction of roughness-induced crossflow-like vorticity in re-entry scenarios,

    F. Ulrich, T. Sedlmeyr, and C. Stemmer, “Methods for neural network based prediction of roughness-induced crossflow-like vorticity in re-entry scenarios,” Int. J. Heat Fluid Flow , vol. 107, p. 109375, Jul. 2024, doi: 10.1016/j.ijheatfluidflow.2024.109375

    work page doi:10.1016/j.ijheatfluidflow.2024.109375 2024

  34. [34]

    Towards Physics-informed Deep Learning for Turbulent Flow Prediction , booktitle =

    “Towards Physics-informed Deep Learning for Turbulent Flow Prediction | Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining.” Accessed: Jul. 25, 2024. [Online]. Available: https://dl.acm.org/doi/abs/10.1145/3394486.3403198

    work page doi:10.1145/3394486.3403198 2024

  35. [35]

    Paleokarst caves recognition from seismic response simulation to convolutional neural network detection,

    D. Zhu et al., “Paleokarst caves recognition from seismic response simulation to convolutional neural network detection,” GEOPHYSICS, vol. 89, no. 1, pp. WA265–WA277, Jan. 2024, doi: 10.1190/geo2023-0133.1

    work page doi:10.1190/geo2023-0133.1 2024

  36. [36]

    FlowDNN: a physics -informed deep neural network for fast and accurate flow prediction | Frontiers of Information Technology & Electronic Engineering

    “FlowDNN: a physics -informed deep neural network for fast and accurate flow prediction | Frontiers of Information Technology & Electronic Engineering.” Accessed: Jul. 25, 2024. [Online]. Available: https://link.springer.com/article/10.1631/FITEE.2000435

    work page doi:10.1631/fitee.2000435 2024

  37. [37]

    Advancing fluid dynamics simulations: A comprehensive approach to optimizing physics -informed neural networks,

    W. Zhou, S. Miwa, and K. Okamoto, “Advancing fluid dynamics simulations: A comprehensive approach to optimizing physics -informed neural networks,” Phys. Fluids, vol. 36, no. 1, Jan. 2024, doi: 10.1063/5.0180770

    work page doi:10.1063/5.0180770 2024

  38. [38]

    Fully Convolutional Networks for Semantic Segmentation,

    J. Long, E. Shelhamer, and T. Darrell, “Fully Convolutional Networks for Semantic Segmentation,” presented at the Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2015, pp. 3431 –3440. Accessed: Jul. 25, 2024. [Online]. Available: https://openaccess.thecvf.com/content_cvpr_2015/html/Long_Fully_Convolutional_Networks _2015_CVP...

    2015

  39. [39]

    In: Medical Image Compu ting and Computer-Assisted Intervention – MICCAI 2015

    O. Ronneberger, P. Fischer, and T. Brox, “U -Net: Convolutional Networks for Biomedical Image Segmentation,” in Medical Image Computing and Computer -Assisted Inter vention – MICCAI 2015, vol. 9351, N. Navab, J. Hornegger, W. M. Wells, and A. F. Frangi, Eds., in Lecture Notes in Computer Science, vol. 9351. , Cham: Springer International Publishing, 2015,...

    work page doi:10.1007/978-3-319-24574-4_28 2015

  40. [40]

    Image -To-Image Translation With Conditional Adversarial Networks,

    P. Isola, J.-Y. Zhu, T. Zhou, and A. A. Efros, “Image -To-Image Translation With Conditional Adversarial Networks,” presented at the Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 1125 –1134. Accessed: Jul. 25, 2024. [Online]. Available: https://openaccess.thecvf.com/content_cvpr_2017/html/Isola_Image -To- Image_T...

    2017

  41. [41]

    Fractal characterization and simulation of rough surfaces,

    A. Majumdar and C. L. Tien, “Fractal characterization and simulation of rough surfaces,” Wear, vol. 136, no. 2, pp. 313–327, Mar. 1990, doi: 10.1016/0043-1648(90)90154-3

    work page doi:10.1016/0043-1648(90)90154-3 1990

  42. [42]

    Exact Spectral Moments and Differentiability of the Weierstrass-Mandelbrot Fractal Function,

    I. Green, “Exact Spectral Moments and Differentiability of the Weierstrass-Mandelbrot Fractal Function,” J. Tribol., vol. 142, no. 4, p. 041501, Apr. 2020, doi: 10.1115/1.4045452

    work page doi:10.1115/1.4045452 2020

  43. [43]

    Study on contact spots of fractal rough surfaces based on three -dimensional Weierstrass -Mandelbrot function,

    J. Chen, F. Yang, K. Luo, Y. Wu, C. Niu, and M. Rong, “Study on contact spots of fractal rough surfaces based on three -dimensional Weierstrass -Mandelbrot function,” in 2016 IEEE 62nd Holm Conference on Electrical Contacts (Holm) , Oct. 2016, pp. 198 –204. doi: 10.1109/HOLM.2016.7780032

    work page doi:10.1109/holm.2016.7780032 2016

  44. [44]

    On the Weierstrass-Mandelbrot Fractal Function,

    M. V. Berry and Z. V. Lewis, “On the Weierstrass-Mandelbrot Fractal Function,” Proc. R. Soc. Lond. Ser. Math. Phys. Sci., vol. 370, no. 1743, pp. 459–484, 1980

    1980

  45. [45]

    A multivariate Weierstrass–Mandelbrot function | Proceedings of the Royal Society of London. A. Mathematical and Physical Sciences

    “A multivariate Weierstrass–Mandelbrot function | Proceedings of the Royal Society of London. A. Mathematical and Physical Sciences.” Accessed: Aug. 05, 2024. [Online]. Available: https://royalsocietypublishing.org/doi/abs/10.1098/rspa.1985.0083

    work page doi:10.1098/rspa.1985.0083 2024

  46. [46]

    How to Generate Random Surfaces in COMSOL Multiphysics® | COMSOL Blog

    “How to Generate Random Surfaces in COMSOL Multiphysics® | COMSOL Blog.” Accessed: Aug. 05, 2024. [Online]. Available: https://www.comsol.com/blogs/how -to-generate-random- surfaces-in-comsol-multiphysics

    2024

  47. [47]

    Ser: Graduate Texts in Mathematics (GTM), vol

    F. Moukalled et al. , “The Finite Volume Method in Computational Fluid Dynamics: An Advanced Introduction with OpenFOAM® and Matlab ,” Aug. 2015, doi: 10.1007/978 -3-319- 16874-6

    work page doi:10.1007/978 2015

  48. [48]

    CBAM: Convolutional Block Attention Module,

    S. Woo, J. Park, J.-Y. Lee, and I. S. Kweon, “CBAM: Convolutional Block Attention Module,” presented at the Proceedings of the European Conference on Computer Vision (ECCV), 2018, pp. 3 –19. Accessed: Jul. 25, 2024. [Online]. Available: https://openaccess.thecvf.com/content_ECCV_2018/html/Sanghyun_Woo_Convolutional_Blo ck_Attention_ECCV_2018_paper.html

    2018

  49. [49]

    T-Net: Nested encoder–decoder architecture for the main vessel segmentation in coronary angiography,

    T. J. Jun, J. Kweon, Y.-H. Kim, and D. Kim, “T-Net: Nested encoder–decoder architecture for the main vessel segmentation in coronary angiography,” Neural Netw., vol. 128, pp. 216 –233, Aug. 2020, doi: 10.1016/j.neunet.2020.05.002

    work page doi:10.1016/j.neunet.2020.05.002 2020

  50. [50]

    PyTorch documentation — PyTorch 2.4 documentation

    “PyTorch documentation — PyTorch 2.4 documentation.” Accessed: Aug. 05, 2024. [Online]. Available: https://pytorch.org/docs/stable/index.html

    2024

  51. [51]

    Deconvolution and Checkerboard Artifacts,

    A. Odena, V. Dumoulin, and C. Olah, “Deconvolution and Checkerboard Artifacts,” Distill, vol. 1, no. 10, p. e3, Oct. 2016, doi: 10.23915/distill.00003

    work page doi:10.23915/distill.00003 2016