Super-resolution analysis via machine learning: a survey for fluid flows
Open accessThe DOI matches its Crossref record; 153 of the 153 references it cites were found in a registry or an index.
Abstract
This paper surveys machine-learning-based super-resolution reconstruction for vortical flows. Super resolution aims to find the high-resolution flow fields from low-resolution data and is generally an approach used in image reconstruction. In addition to surveying a variety of recent super-resolution applications, we provide case studies of super-resolution analysis for an example of two-dimensional decaying isotropic turbulence. We demonstrate that physics-inspired model designs enable successful reconstruction of vortical flows from spatially limited measurements. We also discuss the challenges and outlooks of machine-learning-based super-resolution analysis for fluid flow applications. The insights gained from this study can be leveraged for super-resolution analysis of numerical and experimental flow data. Graphical abstract:
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 8×
“The model is composed of the up-/downsampling operations, the skip connection [91], and CNNs with various sizes of filters.”
BackgroundMethodInfluentialHochreiter, S., & Schmidhuber, J. (1997) · Neural Computation · doi:10.1162/neco.1997.9.8.1735 · Cited by 102,318 · cited 2×
“They reported that transfer learning [92, 139] can help in reducing the required amount of training data, while recurrent neural network (long short-term memory, LSTM [127]) also improves the reconstruction performance of complex transient wake problems.”
BackgroundGradient-based learning applied to document recognition
Lecun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998) · Proceedings of the IEEE · doi:10.1109/5.726791 · Cited by 58,805 · cited 1×
“To address the issue of the computational burden associated with the fully-connected models, convolutional neural networks (CNNs) [76] have been widely utilized in super-resolution analysis of fluid flows.”
Squeeze-and-Excitation Networks
Hu, J., Shen, L., & Sun, G. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00745 · Cited by 35,667
Learning representations by back-propagating errors
Rumelhart, D. E., Hinton, G. E., & Williams, R. J. (1986) · Nature · doi:10.1038/323533a0 · Cited by 31,892
Greedy function approximation: A gradient boosting machine
Friedman, J. H. (2001) · The Annals of Statistics · doi:10.1214/aos/1013203451 · Cited by 30,731 · cited 1×
“With flow examples of two- and three-dimensional cylinder wakes and a spatial mixing layer, they discussed pros and cons of a variety of techniques such as POD [129, 130, 131], regular autoencoder [132], variational autoencoder [133], linear/nonlinear fully-connected networks, support vector machine [134], gradient bo…”
MethodUnpaired image-to-image translation using cycle-consistent adversarial networks
Zhu, J.-Y., Park, T., Isola, P., & Efros, A. A. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.244 · Cited by 23,602
Pan, S. J., & Yang, Q. (2010) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2009.191 · Cited by 23,387 · cited 2×
“SURFNet includes a transfer learning-based augmentation [92].”
BackgroundReducing the dimensionality of data with neural networks
Hinton, G. E., & Salakhutdinov, R. R. (2006) · Science · doi:10.1126/science.1127647 · Cited by 20,928
Raissi, M., Perdikaris, P., & Karniadakis, G. E. (2019) · Journal of Computational Physics · doi:10.1016/j.jcp.2018.10.045 · Cited by 20,384 · cited 4×
“These physics-law-based augmentations inspired by physics-informed neural network (PINN) [84, 85, 140] achieve accurate reconstruction while reducing the required amount of training data [70, 141].”
BackgroundMethodInfluentialGenerative adversarial networks
Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., & Bengio, Y. (2020) · Communications of the ACM · doi:10.1145/3422622 · Cited by 18,384
Nonlinear Dimensionality Reduction by Locally Linear Embedding
Roweis, S. T., & Saul, L. K. (2000) · Science · doi:10.1126/science.290.5500.2323 · Cited by 14,965 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodA tutorial on support vector regression
Smola, A. J., & Schölkopf, B. (2004) · Statistics and Computing · doi:10.1023/b:stco.0000035301.49549.88 · Cited by 13,387 · cited 1×
“With flow examples of two- and three-dimensional cylinder wakes and a spatial mixing layer, they discussed pros and cons of a variety of techniques such as POD [130–132], regular autoencoder [133], variational autoencoder [134], linear/nonlinear fully connected networks, support vector machine [135], gradient boosting…”
MethodPhoto-Realistic Single Image Super-Resolution Using a Generative Adversarial Network
Ledig, C., Theis, L., Huszar, F., Caballero, J., Cunningham, A., Acosta, A., Aitken, A., Tejani, A., Totz, J., Wang, Z., et al. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.19 · Cited by 12,696
Proceedings of the 24th international conference on Machine learning
(2007) · doi:10.1145/1273496 · Cited by 11,630
A Comprehensive Survey on Graph Neural Networks
Wu, Z., Pan, S., Chen, F., Long, G., Zhang, C., & Yu, P. S. (2021) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2020.2978386 · Cited by 11,000 · cited 5×
“To handle spatially irregular sensor arrangements, graph neural network (GNN) [119] can be considered.”
BackgroundInfluentialImage Super-Resolution Using Deep Convolutional Networks
Dong, C., Loy, C. C., He, K., & Tang, X. (2016) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2015.2439281 · Cited by 10,105
Physics-informed machine learning
Karniadakis, G. E., Kevrekidis, I. G., Lu, L., Perdikaris, P., Wang, S., & Yang, L. (2021) · Nature Reviews Physics · doi:10.1038/s42254-021-00314-5 · Cited by 8,718 · cited 1×
“What these terms in the loss function do is to better constrain the solution space [86, 87].”
BackgroundLearning a Deep Convolutional Network for Image Super-Resolution
Dong, C., Loy, C. C., He, K., & Tang, X. (2014) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-10593-2_13 · Cited by 5,494 · cited 1×
“To address this difficulty, machine learning has been used for accurate super-resolution reconstruction of images [26, 27, 28].”
MethodImage Super-Resolution Via Sparse Representation
Yang, J., Wright, J., Huang, T. S., & Ma, Y. (2010) · IEEE Transactions on Image Processing · doi:10.1109/tip.2010.2050625 · Cited by 5,217 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodDiscovering governing equations from data by sparse identification of nonlinear dynamical systems
Brunton, S. L., Proctor, J. L., & Kutz, J. N. (2016) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1517384113 · Cited by 4,796 · cited 1×
“With flow examples of two- and three-dimensional cylinder wakes and a spatial mixing layer, they discussed pros and cons of a variety of techniques such as POD [129, 130, 131], regular autoencoder [132], variational autoencoder [133], linear/nonlinear fully-connected networks, support vector machine [134], gradient bo…”
MethodESRGAN: Enhanced Super-Resolution Generative Adversarial Networks
Wang, X., Yu, K., Wu, S., Gu, J., Liu, Y., Dong, C., Qiao, Y., & Loy, C. C. (2019) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-11021-5_5 · Cited by 4,731
Accelerating the Super-Resolution Convolutional Neural Network
Dong, C., Loy, C. C., & Tang, X. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46475-6_25 · Cited by 3,786 · cited 1×
“To address this difficulty, machine learning has been used for accurate super-resolution reconstruction of images [26, 27, 28].”
Method5th International Conference on Learning Representations (ICLR 17)
Havrylov, S., & Titov, I. (2017) · Cited by 3,503
A few useful things to know about machine learning
Domingos, P. (2012) · Communications of the ACM · doi:10.1145/2347736.2347755 · Cited by 3,326 · cited 1×
“This leads to the number of nodes and their connections to drastically increase, leading to the prohibitively expensive computational cost and the failure of non-convex optimization known as the curse of dimensionality [75].”
BackgroundSuper-resolution image reconstruction: a technical overview
Park, S. C., Park, M. K., & Kang, M. G. (2003) · IEEE Signal Processing Magazine · doi:10.1109/msp.2003.1203207 · Cited by 3,282 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodCubic convolution interpolation for digital image processing
Keys, R. (1981) · IEEE Transactions on Acoustics, Speech, and Signal Processing · doi:10.1109/tassp.1981.1163711 · Cited by 3,255 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodMachine Learning for Fluid Mechanics
Brunton, S. L., Noack, B. R., & Koumoutsakos, P. (2020) · Annual Review of Fluid Mechanics · doi:10.1146/annurev-fluid-010719-060214 · Cited by 3,063
Low-Complexity Single-Image Super-Resolution based on Nonnegative Neighbor Embedding
Bevilacqua, M., Roumy, A., Guillemot, C., & Morel, M.-L. A. (2012) · Procedings of the British Machine Vision Conference 2012 · doi:10.5244/c.26.135 · Cited by 2,701 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodExample-based super-resolution
Freeman, W. T., Jones, T. R., & Pasztor, E. C. (2002) · IEEE Computer Graphics and Applications · doi:10.1109/38.988747 · Cited by 2,498 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodArtificial neural networks for solving ordinary and partial differential equations
Lagaris, I. E., Likas, A., & Fotiadis, D. I. (1998) · IEEE Transactions on Neural Networks · doi:10.1109/72.712178 · Cited by 2,412 · cited 2×
“These physics-law-based augmentations inspired by physics-informed neural network (PINN) [84, 85, 140] achieve accurate reconstruction while reducing the required amount of training data [70, 141].”
BackgroundEfficient sparse coding algorithms
Lee, H., Battle, A., Raina, R., & Ng, A. Y. (2007) · Advances in Neural Information Processing Systems 19 · doi:10.7551/mitpress/7503.003.0105 · Cited by 2,394 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodHidden fluid mechanics: Learning velocity and pressure fields from flow visualizations
Raissi, M., Yazdani, A., & Karniadakis, G. E. (2020) · Science · doi:10.1126/science.aaw4741 · Cited by 2,103
Improving resolution by image registration
Irani, M., & Peleg, S. (1991) · CVGIP: Graphical Models and Image Processing · doi:10.1016/1049-9652(91)90045-l · Cited by 1,929 · cited 2×
“Kai Fukami[1,∗], Koji Fukagata([2]), Kunihiko Taira([1]) 1.”
BackgroundModal Analysis of Fluid Flows: An Overview
Taira, K., Brunton, S. L., Dawson, S. T. M., Rowley, C. W., Colonius, T., McKeon, B. J., Schmidt, O. T., Gordeyev, S., Theofilis, V., & Ukeiley, L. S. (2017) · AIAA Journal · doi:10.2514/1.j056060 · Cited by 1,894 · cited 7×
“With flow examples of two- and three-dimensional cylinder wakes and a spatial mixing layer, they discussed pros and cons of a variety of techniques such as POD [130, 131, 132], regular autoencoder [133], variational autoencoder [134], linear/nonlinear fully-connected networks, support vector machine [135], gradient bo…”
MethodInfluentialSelf-citationSuper-resolution from a single image
Glasner, D., Bagon, S., & Irani, M. (2009) · 2009 IEEE 12th International Conference on Computer Vision · doi:10.1109/iccv.2009.5459271 · Cited by 1,889 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodIntroduction to Semi-Supervised Learning
Zhu, X., & Goldberg, A. B. (2009) · Synthesis Lectures on Artificial Intelligence and Machine Learning · doi:10.1007/978-3-031-01548-9 · Cited by 1,766
Reynolds averaged turbulence modelling using deep neural networks with embedded invariance
Ling, J., Kurzawski, A., & Templeton, J. (2016) · Journal of Fluid Mechanics · doi:10.1017/jfm.2016.615 · Cited by 1,624 · cited 2×
“Ongoing research developments in super-resolution analysis of turbulent flows and data-driven turbulence models [35, 33] may address the issues identified here in the coming years.”
BackgroundTurbulence Modeling in the Age of Data
Duraisamy, K., Iaccarino, G., & Xiao, H. (2019) · Annual Review of Fluid Mechanics · doi:10.1146/annurev-fluid-010518-040547 · Cited by 1,508 · cited 2×
“Ongoing research developments in super-resolution analysis of turbulent flows and data-driven turbulence models [35, 33] may address the issues identified here in the coming years.”
BackgroundFreeman, W. T., Pasztor, E. C., & Carmichael, O. T. (2000) · International Journal of Computer Vision · doi:10.1023/a:1026501619075 · Cited by 1,459 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodTwenty years of particle image velocimetry
Adrian, R. J. (2005) · Experiments in Fluids · doi:10.1007/s00348-005-0991-7 · Cited by 1,267 · cited 1×
“For the preparation of training data in these experimental studies, cross-correlation methods [109] are generally used to obtain velocity fields from particle images.”
MethodLimits on super-resolution and how to break them
Baker, S., & Kanade, T. (2002) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2002.1033210 · Cited by 1,205
Deep Learning for Single Image Super-Resolution: A Brief Review
Yang, W., Zhang, X., Tian, Y., Wang, W., Xue, J.-H., & Liao, Q. (2019) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2019.2919431 · Cited by 1,154 · cited 1×
“To address this difficulty, machine learning has been used for accurate super-resolution reconstruction of images [26, 27, 28].”
MethodFeature Discovery by Competitive Learning*
Rumelhart, D. E., & Zipser, D. (1985) · Cognitive Science · doi:10.1207/s15516709cog0901_5 · Cited by 1,057 · cited 1×
“Summarizing, the parameter inside the generator G is optimized by minimizing the loss function while that for the discriminator D is adjusted by maximizing the loss function, referred to as competitive learning [83].”
MethodMachine learning–accelerated computational fluid dynamics
Kochkov, D., Smith, J. A., Alieva, A., Wang, Q., Brenner, M. P., & Hoyer, S. (2021) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2101784118 · Cited by 1,042 · cited 1×
“Conservatively speaking, turbulent flow statistics may be predicted well with super resolution but highly-accurate reconstruction of each and every instantaneous flow would likely be a major difficulty, if not impossible [172].”
BackgroundTurbulence, Coherent Structures, Dynamical Systems and Symmetry
Holmes, P., Lumley, J. L., Berkooz, G., & Rowley, C. W. (2012) · Cambridge University Press eBooks · doi:10.1017/cbo9780511919701 · Cited by 924
Karhunen–Loève procedure for gappy data
Everson, R., & Sirovich, L. (1995) · Journal of the Optical Society of America A · doi:10.1364/josaa.12.001657 · Cited by 906 · cited 7×
“Similarly, Manohar et al. [126] has also recently performed a fully-connected model and POD-based sparse reconstruction for wake interactions of two cylinders.”
MethodInfluentialSuper-resolution reconstruction of turbulent flows with machine learning
Fukami, K., Fukagata, K., & Taira, K. (2019) · Journal of Fluid Mechanics · doi:10.1017/jfm.2019.238 · Cited by 837 · cited 9×
“For instance, accounting for spatial length scales of the flow structures in the models improves reconstruction [62].”
BackgroundMethodResultInfluentialSelf-citationDiscriminative Unsupervised Feature Learning with Exemplar Convolutional Neural Networks
Dosovitskiy, A., Fischer, P., Springenberg, J. T., Riedmiller, M., & Brox, T. (2016) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2015.2496141 · Cited by 762
Image and video upscaling from local self-examples
Freedman, G., & Fattal, R. (2011) · ACM Transactions on Graphics · doi:10.1145/1944846.1944852 · Cited by 679 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodLi, Y., Perlman, E., Wan, M., Yang, Y., Meneveau, C., Burns, R., Chen, S., Szalay, A., & Eyink, G. (2008) · Journal of Turbulence · doi:10.1080/14685240802376389 · Cited by 673 · cited 1×
“For these reasons, making computational and experimental fluid-flow databases [184, 185, 186] available is critically important to advance studies on data-driven analysis of vortical flows.”
BackgroundGao, H., Sun, L., & Wang, J.-X. (2021) · Journal of Computational Physics · doi:10.1016/j.jcp.2020.110079 · Cited by 621 · cited 1×
“PhyGeoNet [170] includes coordinate transformation from an irregular domain to a structured mesh space for fluid flow regression, allowing us to convolve on flow fields, as illustrated in figure 13(c).”
MethodSingle Image Super-Resolution With Non-Local Means and Steering Kernel Regression
Zhang, K., Gao, X., Tao, D., & Li, X. (2012) · IEEE Transactions on Image Processing · doi:10.1109/tip.2012.2208977 · Cited by 576 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodRabault, J., Kuchta, M., Jensen, A., Réglade, U., & Cerardi, N. (2019) · Journal of Fluid Mechanics · doi:10.1017/jfm.2019.62 · Cited by 575 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundStolz, S., Adams, N. A., & Kleiser, L. (2001) · Physics of Fluids · doi:10.1063/1.1350896 · Cited by 574 · cited 1×
“Similar to an approximate deconvolution model [173] which considers inverse mapping of spatial filters, super resolution could be used to augment the subgrid-scale models.”
MethodPSF estimation using sharp edge prediction
Joshi, N., Szeliski, R., & Kriegman, D. J. (2008) · 2008 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2008.4587834 · Cited by 563 · cited 2×
“For training the baseline networks, we use 1000 snapshots over an eddy turn-overtime of t ∈ [2, 6] with a time interval of ∆t = 0.”
Method(2018) · IEEE Control Systems · doi:10.1109/mcs.2018.2810460 · Cited by 518 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundDNS-based predictive control of turbulence: an optimal benchmark for feedback algorithms
BEWLEY, T. R., MOIN, P., & TEMAM, R. (2001) · Journal of Fluid Mechanics · doi:10.1017/s0022112001005821 · Cited by 452
A direct numerical simulation study on the mean velocity characteristics in turbulent pipe flow
WU, X., & MOIN, P. (2008) · Journal of Fluid Mechanics · doi:10.1017/s0022112008002085 · Cited by 421 · cited 1×
“For these reasons, making computational and experimental fluid-flow databases [184, 185, 186] available is critically important to advance studies on data-driven analysis of vortical flows.”
BackgroundDeep learning methods for super-resolution reconstruction of turbulent flows
Liu, B., Tang, J., Huang, H., & Lu, X.-Y. (2020) · Physics of Fluids · doi:10.1063/1.5140772 · Cited by 359 · cited 2×
“[66] compared two types of supervised CNN-based models for super-resolution analysis: namely the static CNN (SCNN) and the multiple temporal paths CNN (MTPC).”
BackgroundMethodNonlinear mode decomposition with convolutional neural networks for fluid dynamics
Murata, T., Fukami, K., & Fukagata, K. (2020) · Journal of Fluid Mechanics · doi:10.1017/jfm.2019.822 · Cited by 359
Self-citationNonparametric Blind Super-resolution
Michaeli, T., & Irani, M. (2013) · 2013 IEEE International Conference on Computer Vision · doi:10.1109/iccv.2013.121 · Cited by 332 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodFast Image Super-Resolution Based on In-Place Example Regression
Yang, J., Lin, Z., & Cohen, S. (2013) · 2013 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2013.141 · Cited by 327 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodData-driven prediction of unsteady flow over a circular cylinder using deep learning
Lee, S., & You, D. (2019) · Journal of Fluid Mechanics · doi:10.1017/jfm.2019.700 · Cited by 327
Machine-learning-based spatio-temporal super resolution reconstruction of turbulent flows
Fukami, K., Fukagata, K., & Taira, K. (2021) · Journal of Fluid Mechanics · doi:10.1017/jfm.2020.948 · Cited by 306 · cited 3×
“We note that our previous studies [62, 90] on machine-learning-based super-resolution reconstruction was performed with two-dimensional decaying turbulence but at lower Reynolds numbers (Re0 ≈ 80) with smaller numbers of the grid points (N = 128).”
BackgroundMethodResultSelf-citationA point-cloud deep learning framework for prediction of fluid flow fields on irregular geometries
Kashefi, A., Rempe, D., & Guibas, L. J. (2021) · Physics of Fluids · doi:10.1063/5.0033376 · Cited by 295 · cited 1×
“[166] has recently applied it to fluid flows.”
BackgroundPerspective on machine learning for advancing fluid mechanics
Brenner, M. P., Eldredge, J. D., & Freund, J. B. (2019) · Physical Review Fluids · doi:10.1103/physrevfluids.4.100501 · Cited by 292 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundAdversarial super-resolution of climatological wind and solar data
Stengel, K., Glaws, A., Hettinger, D., & King, R. N. (2020) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1918964117 · Cited by 287 · cited 1×
“[158] has comprehensively compared various models for super-resolution reconstruction, including a super-resolution GAN [159], stochastic estimation, a deconvolution GAN [160], and diversity-sensitive conditional GAN [161].”
BackgroundUnsupervised deep learning for super-resolution reconstruction of turbulence
Kim, H., Kim, J., Won, S., & Lee, C. (2021) · Journal of Fluid Mechanics · doi:10.1017/jfm.2020.1028 · Cited by 285 · cited 4×
“[67], the unsupervised GAN-based super resolution has recently been examined for a variety of flows.”
BackgroundInfluentialRAISR: Rapid and Accurate Image Super Resolution
Romano, Y., Isidoro, J., & Milanfar, P. (2017) · IEEE Transactions on Computational Imaging · doi:10.1109/tci.2016.2629284 · Cited by 280 · cited 1×
“One can also implement a resize or interpolation function for the input data to align the size with that of the output [29,54,80].”
Deng, Z., He, C., Liu, Y., & Kim, K. C. (2019) · Physics of Fluids · doi:10.1063/1.5127031 · Cited by 275 · cited 1×
“[106] developed a machine-learning model to super-resolve PIV measurements.”
Predictions of turbulent shear flows using deep neural networks
Srinivasan, P. A., Guastoni, L., Azizpour, H., Schlatter, P., & Vinuesa, R. (2019) · Physical Review Fluids · doi:10.1103/physrevfluids.4.054603 · Cited by 274 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundShallow neural networks for fluid flow reconstruction with limited sensors
Erichson, N. B., Mathelin, L., Yao, Z., Brunton, S. L., Mahoney, M. W., & Kutz, J. N. (2020) · Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences · doi:10.1098/rspa.2020.0097 · Cited by 259 · cited 4×
“Hence, models generally comprise the decoder-type structure [63, 74], meaning that the number of nodes gradually increases towards the output layer.”
BackgroundMethodInfluentialGlobal field reconstruction from sparse sensors with Voronoi tessellation-assisted deep learning
Fukami, K., Maulik, R., Ramachandra, N., Fukagata, K., & Taira, K. (2021) · Nature Machine Intelligence · doi:10.1038/s42256-021-00402-2 · Cited by 253 · cited 4×
“Pradhan, Duraisamy (2021) [59] Turbulent channel flow Supervised MLP (VSRNN) Generalized for unseen initial condition, expensive Fukami, Maulik, Ramachandra, Fukagata, Taira (2021) [53] Laminar cylinder wake, sea surface temperature, turbulent channel flow Supervised CNN (Voronoi model) Generalized for moving sensors,…”
BackgroundMethodInfluentialSelf-citationAssessment of supervised machine learning methods for fluid flows
Fukami, K., Fukagata, K., & Taira, K. (2020) · Theoretical and Computational Fluid Dynamics · doi:10.1007/s00162-020-00518-y · Cited by 234
Self-citationPhysics-constrained bayesian neural network for fluid flow reconstruction with sparse and noisy data
Sun, L., & Wang, J.-X. (2020) · Theoretical and Applied Mechanics Letters · doi:10.1016/j.taml.2020.01.031 · Cited by 223 · cited 1×
“In this aspect, the inverse problem of global field reconstruction from local measurements is an extension of super-resolution analysis [53, 54, 55].”
MethodDense motion estimation of particle images via a convolutional neural network
Cai, S., Zhou, S., Xu, C., & Gao, Q. (2019) · Experiments in Fluids · doi:10.1007/s00348-019-2717-2 · Cited by 216 · cited 1×
“[110] used FlowNetS [111] to estimate velocity fields of cylinder wake, backward-facing step flow, and isotropic turbulence from synthetic particle images.”
MethodConvolutional-network models to predict wall-bounded turbulence from wall quantities
Guastoni, L., Güemes, A., Ianiro, A., Discetti, S., Schlatter, P., Azizpour, H., & Vinuesa, R. (2021) · Journal of Fluid Mechanics · doi:10.1017/jfm.2021.812 · Cited by 209 · cited 1×
“Transfer learning over multiple levels of spatial-resolution flow field can improve the accuracy of super-resolution reconstruction [93], which is also related to multi-fidelity learning [94].”
BackgroundFukami, K., Nakamura, T., & Fukagata, K. (2020) · Physics of Fluids · doi:10.1063/5.0020721 · Cited by 205 · cited 1×
“By visualizing the weight distribution between the latent space representation and the whole field, the shallow decoder provides nonlinear modes that represent the contribution of each latent variable for super-resolution reconstruction, which are analogous to those captured by nonlinear autoencoders [108, 114, 115, 1…”
BackgroundSelf-citationMeta-learning PINN loss functions
Psaros, A. F., Kawaguchi, K., & Karniadakis, G. E. (2022) · Journal of Computational Physics · doi:10.1016/j.jcp.2022.111121 · Cited by 196 · cited 1×
“can consider the use of optimization for finding an optimal set of coefficients, although it is computationally expensive [148].”
BackgroundHasegawa, K., Fukami, K., Murata, T., & Fukagata, K. (2020) · Theoretical and Computational Fluid Dynamics · doi:10.1007/s00162-020-00528-w · Cited by 195 · cited 1×
“Weighting the gradient of the flowfield promotes a smooth and physically plausible reconstruction [146,147].”
BackgroundSelf-citationPrediction of turbulent heat transfer using convolutional neural networks
Kim, J., & Lee, C. (2020) · Journal of Fluid Mechanics · doi:10.1017/jfm.2019.814 · Cited by 194 · cited 1×
“Kim and Lee [100] considered a CNN-based estimation of the high-resolution heat flux field in a turbulent channel flow from poorly-resolved wall-shear stresses and pressure.”
BackgroundGao, H., Sun, L., & Wang, J.-X. (2021) · Physics of Fluids · doi:10.1063/5.0054312 · Cited by 193 · cited 6×
“[56] proposed a semisupervised CNN-based super-resolution analysis for fluid flows.”
BackgroundMethodResultInfluentialAutomating turbulence modelling by multi-agent reinforcement learning
Novati, G., de Laroussilhe, H. L., & Koumoutsakos, P. (2021) · Nature Machine Intelligence · doi:10.1038/s42256-020-00272-0 · Cited by 182 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundSparse identification of nonlinear dynamics with low-dimensionalized flow representations
Fukami, K., Murata, T., Zhang, K., & Fukagata, K. (2021) · Journal of Fluid Mechanics · doi:10.1017/jfm.2021.697 · Cited by 182
Self-citationRobust flow reconstruction from limited measurements via sparse representation
Callaham, J. L., Maeda, K., & Brunton, S. L. (2019) · Physical Review Fluids · doi:10.1103/physrevfluids.4.103907 · Cited by 178 · cited 1×
“With flow examples of two- and three-dimensional cylinder wakes and a spatial mixing layer, they discussed pros and cons of a variety of techniques such as POD [130–132], regular autoencoder [133], variational autoencoder [134], linear/nonlinear fully connected networks, support vector machine [135], gradient boosting…”
MethodTowards extraction of orthogonal and parsimonious non-linear modes from turbulent flows
Eivazi, H., Clainche, S. L., Hoyas, S., & Vinuesa, R. (2022) · Expert Systems with Applications · doi:10.1016/j.eswa.2022.117038 · Cited by 175
Scientific multi-agent reinforcement learning for wall-models of turbulent flows
Bae, H. J., & Koumoutsakos, P. (2022) · Nature Communications · doi:10.1038/s41467-022-28957-7 · Cited by 171 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundBode, M., Gauding, M., Lian, Z., Denker, D., Davidovic, M., Kleinheinz, K., Jitsev, J., & Pitsch, H. G. (2021) · Proceedings of the Combustion Institute · doi:10.1016/j.proci.2020.06.022 · Cited by 168 · cited 4×
“Yousif, Yu, Lim (2021) [70] Turbulent channel flow Unsupervised CNN (MS-ESRGAN) Robust for multi-scale physics, expensive Bode, Gauding, Lian, Denker, Davidovic, Kleinheinz, Jitsev, Pitsch (2021) [60] Isotropic turbulence Unsupervised CNN (PIESRGAN) Physics loss function, tuning of loss coefficients”
BackgroundInfluentialHasegawa, K., Fukami, K., Murata, T., & Fukagata, K. (2020) · Fluid Dynamics Research · doi:10.1088/1873-7005/abb91d · Cited by 166
Self-citationNeural network closures for nonlinear model order reduction
San, O., & Maulik, R. (2018) · Advances in Computational Mathematics · doi:10.1007/s10444-018-9590-z · Cited by 165 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundProbabilistic neural networks for fluid flow surrogate modeling and data recovery
Maulik, R., Fukami, K., Ramachandra, N., Fukagata, K., & Taira, K. (2020) · Physical Review Fluids · doi:10.1103/physrevfluids.5.104401 · Cited by 164 · cited 1×
“To address this issue, there are also some efforts to estimate low-order representations such as coefficients obtained through proper orthogonal decomposition (POD) from sparse sensor measurements [121, 122, 123].”
BackgroundSelf-citationRobust flow control and optimal sensor placement using deep reinforcement learning
Paris, R., Beneddine, S., & Dandois, J. (2021) · Journal of Fluid Mechanics · doi:10.1017/jfm.2020.1170 · Cited by 162 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundFrom coarse wall measurements to turbulent velocity fields through deep learning
Güemes, A., Discetti, S., Ianiro, A., Sırmaçek, B., Azizpour, H., & Vinuesa, R. (2021) · Physics of Fluids · doi:10.1063/5.0058346 · Cited by 154 · cited 3×
“In addition to supervised fully-connected networks and convolutional networks, unsupervised learning with generative adversarial network (GAN) [82] has also been proposed for super-resolution analysis of fluid flows [61, 64, 67, 69].”
BackgroundMethodGrasping extreme aerodynamics on a low-dimensional manifold
Fukami, K., & Taira, K. (2023) · Nature Communications · doi:10.1038/s41467-023-42213-6 · Cited by 153 · cited 1×
“By visualizing the weight distribution between the latent space representation and the whole field, the shallow decoder provides nonlinear modes that represent the contribution of each latent variable for super-resolution reconstruction, which are analogous to those captured by nonlinear autoencoders [108, 114, 115, 1…”
BackgroundSelf-citationMorimoto, M., Fukami, K., Zhang, K., Nair, A., & Fukagata, K. (2021) · Theoretical and Computational Fluid Dynamics · doi:10.1007/s00162-021-00580-0 · Cited by 137 · cited 1×
“CNNs incorporate a function called filter sharing, enabling the processing of large vortical flow data without encountering the curse of dimensionality [77].”
BackgroundSelf-citationRobust principal component analysis for modal decomposition of corrupt fluid flows
Scherl, I., Strom, B., Shang, J. K., Williams, O. J. H., Polagye, B. L., & Brunton, S. L. (2020) · Physical Review Fluids · doi:10.1103/physrevfluids.5.054401 · Cited by 132 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundNetwork structure of two-dimensional decaying isotropic turbulence
Taira, K., Nair, A. G., & Brunton, S. L. (2016) · Journal of Fluid Mechanics · doi:10.1017/jfm.2016.235 · Cited by 131 · cited 9×
“They also demonstrated that the model trained with data from DNS can be applied to the LES data.”
MethodInfluentialSelf-citationMachine learning for fluid flow reconstruction from limited measurements
Dubois, P., Gomez, T., Planckaert, L., & Perret, L. (2021) · Journal of Computational Physics · doi:10.1016/j.jcp.2021.110733 · Cited by 128 · cited 1×
“Figure 6: Reduced-order modeling-assisted super-resolution reconstruction [71, 128].”
Deep model predictive flow control with limited sensor data and online learning
Bieker, K., Peitz, S., Brunton, S. L., Kutz, J. N., & Dellnitz, M. (2020) · Theoretical and Computational Fluid Dynamics · doi:10.1007/s00162-020-00520-4 · Cited by 123 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundYousif, M. Z., Yu, L., & Lim, H.-C. (2021) · Physics of Fluids · doi:10.1063/5.0066077 · Cited by 122 · cited 3×
“These physics-law-based augmentations inspired by physics-informed neural network (PINN) [84, 85, 140] achieve accurate reconstruction while reducing the required amount of training data [70, 141].”
BackgroundDeep learning to discover and predict dynamics on an inertial manifold
Linot, A. J., & Graham, M. D. (2020) · Physical review. E · doi:10.1103/physreve.101.062209 · Cited by 122
Space-time super-resolution from a single video
Shahar, O., Faktor, A., & Irani, M. (2011) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2011.5995360 · Cited by 121 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodSuper-resolution and denoising of 4D-Flow MRI using physics-Informed deep neural nets
Fathi, M. F., Perez‐Raya, I., Baghaie, A., Berg, P., Janiga, G., Arzani, A., & D’Souza, R. M. (2020) · Computer Methods and Programs in Biomedicine · doi:10.1016/j.cmpb.2020.105729 · Cited by 117 · cited 1×
“If we consider low-resolution fluid flow data as noisy experimental measurements, super-resolution analysis can also be extended to denoising problem [56, 57, 58].”
MethodSensing coherent structures from the wall
Güemes, A. (2021) · e-Archivo (Carlos III University of Madrid) · Cited by 116
State estimation in wall-bounded flow systems. Part 3. The ensemble Kalman filter
Colburn, C., Cessna, J., & Bewley, T. (2011) · Journal of Fluid Mechanics · doi:10.1017/jfm.2011.222 · Cited by 116 · cited 1×
“[69] combined a GAN-based super-resolution reconstruction and state estimation [152, 153, 154, 155] from the wall sensor measurements of turbulent channel flow.”
MethodA deep-learning approach for reconstructing 3D turbulent flows from 2D observation data
Yousif, M. Z., Yu, L., Hoyas, S., Vinuesa, R., & Lim, H.-C. (2023) · Scientific Reports · doi:10.1038/s41598-023-29525-9 · Cited by 110 · cited 1×
“These physics-law-based augmentations inspired by physics-informed neural network (PINN) [84, 85, 140] achieve accurate reconstruction while reducing the required amount of training data [70, 141].”
BackgroundSeparating Signal from Noise Using Patch Recurrence across Scales
Zontak, M., Mosseri, I., & Irani, M. (2013) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2013.158 · Cited by 108 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodExperimental velocity data estimation for imperfect particle images using machine learning
Morimoto, M., Fukami, K., & Fukagata, K. (2021) · Physics of Fluids · doi:10.1063/5.0060760 · Cited by 105 · cited 1×
“The super-resolution approach with particle images has also been applied to a wake around bluff bodies to remove the influence of reflection and halation in PIV measurements [113].”
MethodSelf-citationGeometry constrained sparse coding for single image super-resolution
Lu, X., Yuan, H., Yan, P., Yuan, Y., & Li, X. (2012) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2012.6247858 · Cited by 104 · cited 1×
“This problem set has been traditionally tackled in computer visions with various techniques including interpolation [4, 5, 6, 7], example-based internal learning [8, 9, 10, 11], high-frequency transfer [12, 13, 14, 15], neighbor embedding [16, 17, 18, 19, 20], and sparse coding [21, 22, 23, 24, 25].”
MethodGeneralization techniques of neural networks for fluid flow estimation
Morimoto, M., Fukami, K., Zhang, K., & Fukagata, K. (2021) · Neural Computing and Applications · doi:10.1007/s00521-021-06633-z · Cited by 104 · cited 1×
“[101] has recently examined the effect of inter- and extrapolation with machine-learning-based super-resolution reconstruction with respect to flow parameters.”
BackgroundSelf-citationPhysics-guided deep learning for generating turbulent inflow conditions
Yousif, M. Z., Yu, L., & Lim, H.-C. (2022) · Journal of Fluid Mechanics · doi:10.1017/jfm.2022.61 · Cited by 95 · cited 1×
“For instance, the use of multi-scale filters inside CNN can be generalized not only in supervised learning but also in unsupervised techniques [174].”
BackgroundSub-grid scale model classification and blending through deep learning
Maulik, R., San, O., Jacob, J., & Crick, C. (2019) · Journal of Fluid Mechanics · doi:10.1017/jfm.2019.254 · Cited by 95 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundSingle-step deep reinforcement learning for open-loop control of laminar and turbulent flows
Ghraieb, H., Viquerat, J., Larcher, A., Méliga, P., & Hachem, E. (2021) · Physical Review Fluids · doi:10.1103/physrevfluids.6.053902 · Cited by 91
A comparison of neural network architectures for data-driven reduced-order modeling
Gruber, A., Gunzburger, M., Ju, L., & Wang, Z. (2022) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2022.114764 · Cited by 89
PhySR: Physics-informed deep super-resolution for spatiotemporal data
Ren, P., Rao, C., Liu, Y., Ma, Z., Wang, Q., Wang, J., & Sun, H. (2023) · Journal of Computational Physics · doi:10.1016/j.jcp.2023.112438 · Cited by 88
A Database for Reduced-Complexity Modeling of Fluid Flows
Towne, A., Dawson, S. T. M., Brès, G. A., Lozano-Durán, A., Saxton-Fox, T., Parthasarathy, A., Jones, A. R., Biler, H., Yeh, C.-A., Patel, H. D., et al. (2023) · AIAA Journal · doi:10.2514/1.j062203 · Cited by 85 · cited 1×
“For these reasons, making computational and experimental fluid-flow databases [184, 185, 186] available is critically important to advance studies on data-driven analysis of vortical flows.”
BackgroundSelf-citationSuper-resolution generative adversarial networks of randomly-seeded fields
Güemes, A., Vila, C. S., & Discetti, S. (2022) · Nature Machine Intelligence · doi:10.1038/s42256-022-00572-7 · Cited by 81
Yousif, M. Z., Yu, L., & Lim, H.-C. (2022) · Physics of Fluids · doi:10.1063/5.0074724 · Cited by 76 · cited 1×
“[156] extended a super-resolution GAN model by combining it with multi-scale CNN [62] and applied it to a turbulent channel flow with large longitudinal ribs.”
MethodSpecial issue on machine learning and data-driven methods in fluid dynamics
Brunton, S. L., Hemati, M. S., & Taira, K. (2020) · Theoretical and Computational Fluid Dynamics · doi:10.1007/s00162-020-00542-y · Cited by 75 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundSelf-citationLeoni, P. C. D., Mazzino, A., & Biferale, L. (2018) · Physical Review Fluids · doi:10.1103/physrevfluids.3.104604 · Cited by 69
Machine-learning-based feedback control for drag reduction in a turbulent channel flow
Park, J. H., & Choi, H. (2020) · Journal of Fluid Mechanics · doi:10.1017/jfm.2020.690 · Cited by 68 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundLeveraging reduced-order models for state estimation using deep learning
Nair, N. J., & Goza, A. J. (2020) · Journal of Fluid Mechanics · doi:10.1017/jfm.2020.409 · Cited by 67 · cited 3×
“Nair, Goza (2020) [71] Laminar flat plate wake Supervised MLP Inexpensive, requires POD”
BackgroundKong, C., Chang, J., Li, Y., & Chen, R. (2020) · AIP Advances · doi:10.1063/5.0030040 · Cited by 65
State estimation in minimal turbulent channel flow: A comparative study of 4DVar and PINN
Du, Y., Wang, M., & Zaki, T. A. (2022) · International Journal of Heat and Fluid Flow · doi:10.1016/j.ijheatfluidflow.2022.109073 · Cited by 60 · cited 1×
“Since the majority of real-world problems do not have access to ground truth and only sparse and noisy measurements are available, one can consider the use of data assimilation [177, 178, 179] to improve super-resolution reconstruction by incorporating the latest observations with a short-range real-time forecast.”
BackgroundSuzuki, T., & Hasegawa, Y. (2017) · Journal of Fluid Mechanics · doi:10.1017/jfm.2017.580 · Cited by 58
Morimoto, M., Fukami, K., Maulik, R., Vinuesa, R., & Fukagata, K. (2022) · Physica D Nonlinear Phenomena · doi:10.1016/j.physd.2022.133454 · Cited by 56
Self-citationWang, H., Yang, Z., Li, B., & Wang, S. (2020) · Physics of Fluids · doi:10.1063/5.0023786 · Cited by 56 · cited 1×
“[107] proposed a super-resolution neural network for two-dimensional PIV (PIV2DSR) based on CNNs.”
BackgroundMultiresolution convolutional autoencoders
Liu, Y., Ponce, C., Brunton, S. L., & Kutz, J. N. (2022) · Journal of Computational Physics · doi:10.1016/j.jcp.2022.111801 · Cited by 55
Sparse sensor reconstruction of vortex-impinged airfoil wake with machine learning
Zhong, Y., Fukami, K., An, B., & Taira, K. (2023) · Theoretical and Computational Fluid Dynamics · doi:10.1007/s00162-023-00657-y · Cited by 53 · cited 1×
“[140, 141] for a vortex-airfoil gust interaction problem.”
Self-citationAn empirical mean-field model of symmetry-breaking in a turbulent wake
Callaham, J., Rigas, G., Loiseau, J.-C., & Brunton, S. L. (2022) · Science Advances · doi:10.1126/sciadv.abm4786 · Cited by 53 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundSuper-Resolution Simulation for Real-Time Prediction of Urban Micrometeorology
Onishi, R., Sugiyama, D., & Matsuda, K. (2019) · SOLA · doi:10.2151/sola.2019-032 · Cited by 53 · cited 2×
“[102] proposed a CNN-based model for super-resolution analysis of temperature fields in urban environment.”
MethodMachine learning for flow field measurements: a perspective
Discetti, S., & Liu, Y. (2022) · Measurement Science and Technology · doi:10.1088/1361-6501/ac9991 · Cited by 47 · cited 1×
“In addition to the aforementioned studies with numerical data, applications to experiments have also been considered [105].”
BackgroundYasuda, Y., Onishi, R., Hirokawa, Y., Kolomenskiy, D., & Sugiyama, D. (2021) · Building and Environment · doi:10.1016/j.buildenv.2021.108597 · Cited by 46 · cited 2×
“Supervised super-resolution techniques have also been applied to larger-scale meteorological flows [102, 103].”
BackgroundMethodNeural network–based pore flow field prediction in porous media using super resolution
Zhou, X., McClure, J., Chen, C., et al. (2022) · Physical Review Fluids · doi:10.1103/physrevfluids.7.074302 · Cited by 45 · cited 1×
“[61] Porous flow Supervised CNN (U-Net) Robust against noise, need to compress data Güemes et al.”
MethodExample-Based Super Resolution
(2016) · Elsevier eBooks · doi:10.1016/c2015-0-06719-3 · Cited by 42
Predicting drag on rough surfaces by transfer learning of empirical correlations
Lee, S.-S., Yang, J., Forooghi, P., Stroh, A., & Bagheri, S. (2021) · Journal of Fluid Mechanics · doi:10.1017/jfm.2021.1041 · Cited by 39 · cited 1×
“They reported that transfer learning [92, 139] can help in reducing the required amount of training data, while recurrent neural network (long short-term memory, LSTM [127]) also improves the reconstruction performance of complex transient wake problems.”
BackgroundData-driven sparse reconstruction of flow over a stalled aerofoil using experimental data
Carter, D., Voogt, F. D., Soares, R. F., & Ganapathisubramani, B. (2021) · Data-Centric Engineering · doi:10.1017/dce.2021.5 · Cited by 39 · cited 1×
“To address this issue, there are also some efforts to estimate low-order representations such as coefficients obtained through proper orthogonal decomposition (POD) from sparse sensor measurements [120, 121, 122].”
BackgroundGiannopoulos, A., & Aider, J. (2020) · Physics of Fluids · doi:10.1063/5.0015870 · Cited by 39
Deep Learning at Scale for Subgrid Modeling in Turbulent Flows: Regression and Reconstruction
Bode, M., Gauding, M., Kleinheinz, K., & Pitsch, H. G. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-34356-9_41 · Cited by 38
Nakamura, T., Fukami, K., & Fukagata, K. (2022) · Scientific Reports · doi:10.1038/s41598-022-07515-7 · Cited by 36 · cited 1×
“However, it is necessary to assess the robustness and sensitivity of the models against noisy inputs [175].”
BackgroundSelf-citationSURFNet: Super-Resolution of Turbulent Flows with Transfer Learning using Small Datasets
Obiols-Sales, O., Vishnu, A., Malaya, N. P., & Chandramowlishwaran, A. M. (2021) · International Conference on Parallel Architectures and Compilation Techniques (PACT) · doi:10.1109/pact52795.2021.00031 · Cited by 34 · cited 6×
“Analogous to SURFNet [65], a hierarchical GAN is first trained with low-resolution data sets.”
BackgroundMethodInfluentialFukami, K., An, B., Nohmi, M., Obuchi, M., & Taira, K. (2022) · Journal of Fluids Engineering · doi:10.1115/1.4055178 · Cited by 31 · cited 1×
“This point will be important as machine-learning-based super-resolution analyses become utilized in industrial applications [176].”
BackgroundSelf-citationDeep Hierarchical Super Resolution for Scientific Data
Wurster, S. W., Guo, H., Shen, H., Peterka, T., & Xu, J. (2022) · IEEE Transactions on Visualization and Computer Graphics · doi:10.1109/tvcg.2022.3214420 · Cited by 29 · cited 1×
“For example, the upsampling can be used inside a network to expand the dimension [79].”
BackgroundZhang, B., Ooka, R., Kikumoto, H., Hu, C., & Tse, K. T. (2022) · Journal of Wind Engineering and Industrial Aerodynamics · doi:10.1016/j.jweia.2022.105243 · Cited by 29
Adversarial sampling of unknown and high-dimensional conditional distributions
Hassanaly, M., Glaws, A., Stengel, K., & King, R. (2021) · Journal of Computational Physics · doi:10.1016/j.jcp.2021.110853 · Cited by 29 · cited 1×
“[158] has comprehensively compared various models for super-resolution reconstruction, including a super-resolution GAN [159], stochastic estimation, a deconvolution GAN [160], and diversity-sensitive conditional GAN [161].”
BackgroundWang, X., Zhu, S., Guo, Y., Han, P., Wang, Y., Wei, Z., & Jin, X. (2022) · Journal of Computational Science · doi:10.1016/j.jocs.2022.101906 · Cited by 28 · cited 4×
“[146] have recently proposed TransFlowNet which weakens the constraint of initial and boundary conditions compared to MeshfreeFlowNet.”
MethodResultInfluentialInterpretable deep learning for prediction of Prandtl number effect in turbulent heat transfer
Kim, H., Kim, J., & Lee, C. (2023) · Journal of Fluid Mechanics · doi:10.1017/jfm.2022.1069 · Cited by 28 · cited 1×
“Machine-learning-based techniques in general [30, 31, 32] have been considered for a range of applications in fluid mechanics including turbulence modeling [33, 34, 35, 36, 37], reduced-order modeling [38, 39, 40, 41, 42], data reconstruction [43, 44, 45, 46], and flow control [47, 48, 49, 50, 51, 52].”
BackgroundRobust training approach of neural networks for fluid flow state estimations
Nakamura, T., & Fukagata, K. (2022) · International Journal of Heat and Fluid Flow · doi:10.1016/j.ijheatfluidflow.2022.108997 · Cited by 23 · cited 1×
“For regression tasks, it is useful for reducing spatial sensitivity, producing a robust CNN model against noisy inputs [78].”
MethodSelf-citationYasuda, Y., & Onishi, R. (2023) · Journal of Advances in Modeling Earth Systems · doi:10.1029/2023ms003658 · Cited by 20 · cited 2×
“Yasuda and Onishi [180] has recently proposed a four-dimensional super-resolution data assimilation and demonstrated its performance with a two-dimensional periodic channel flow.”
BackgroundSparse sensor-based cylinder flow estimation using artificial neural networks
Manohar, K. H., Morton, C. R., & Ziadé, P. (2022) · Physical Review Fluids · doi:10.1103/physrevfluids.7.024707 · Cited by 15 · cited 1×
“[128] has also recently performed a fully connectedmodel andPOD-based sparse reconstruction forwake interactions of two cylinders.”
BackgroundPradhan, A., & Duraisamy, K. (2021) · SSRN Electronic Journal · doi:10.2139/ssrn.3991780 · Cited by 3
Machine-learning-based reconstruction of transient vortex-airfoil wake interaction
Zhong, Y., Fukami, K., An, B., & Taira, K. (2022) · AIAA AVIATION Forum · doi:10.2514/6.2022-3244 · Cited by 2 · cited 2×
“For example, the upsampling can be used inside a network to expand the dimension [79, 80].”
BackgroundSelf-citation
Metadata
- Authors
- Fukami, K., Fukagata, K., & Taira, K.
- Venue
- Theoretical and Computational Fluid Dynamics
- Published
- 2023-06-16
- Volume
- 37
- Issue
- 4
- Pages
- 421-444
- ISSN
- 0935-4964 (ISSN-L)
- 1432-2250
- OpenAlex
- W4380989065
- Cited by
- 176
- References
- 153
- FWCI
- 17.79
- Type
- journal article
- Language
- en
- Open access
- hybrid
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.