×

Deep learning for physical processes: incorporating prior scientific knowledge. (English) Zbl 1459.68183

Summary: We consider the use of deep learning methods for modeling complex phenomena like those occurring in natural physical processes. With the large amount of data gathered on these phenomena the data intensive paradigm could begin to challenge more traditional approaches elaborated over the years in fields like maths or physics. However, despite considerable successes in a variety of application domains, the machine learning field is not yet ready to handle the level of complexity required by such problems. Using an example application, namely sea surface temperature prediction, we show how general background knowledge gained from the physics could be used as a guideline for designing efficient deep learning models. In order to motivate the approach and to assess its generality we demonstrate a formal link between the solution of a class of differential equations underlying a large family of physical phenomena and the proposed model. Experiments and comparison with series of baselines including a state of the art numerical approach is then provided.

MSC:

68T07 Artificial neural networks and deep learning
86A05 Hydrology, hydrography, oceanography

Software:

FlowNet
PDFBibTeX XMLCite
Full Text: DOI arXiv

References:

[1] Álvarez M A, Luengo D and Lawrence N D 2013 Linear latent force models using gaussian processes IEEE Transactions on Pattern Analysis and Machine Intelligence 35 2693-705 · doi:10.1109/TPAMI.2013.86
[2] Archambeau C, Cornford D, Opper M and Shawe-Taylor J 2007 Gaussian process approximations of stochastic differential equations Gaussian Processes in Practice pp 1-16
[3] Béréziat D and Herlin I 2015 Coupling Dynamic Equations and Satellite Images for Modelling Ocean Surface Circulation (Cham: Springer) pp 191-205
[4] Bernstein R L 1982 Sea surface temperature estimation using the NOAA 6 satellite advanced very high resolution radiometer J. Geophys. Res.: Oceans87 9455-65 · doi:10.1029/JC087iC12p09455
[5] Brajard J, Charantonis A A and Jourdin F 2017 Predicting ocean dynamics through machine learning: application on sea-surface suspended particulate mater American Meteorological Society Annual Meeting 2017
[6] Chung J, Kastner K, Dinh L, Goel K, Courville A C and Bengio Y 2015 A recurrent latent variable model for sequential data Advances in Neural Information Processing Systems 28 ed C Cortes et al (Newport, RI: Curran Associates, Inc.) pp 2980-8
[7] Cressie N and Wikle C K 2015 Statistics for Spatio-Temporal Data (New York: Wiley)
[8] Finn C, Goodfellow I J and Levine S 2016 Unsupervised learning for physical interaction through video prediction CoRR (arXiv:1605.07157)
[9] Fischer P, Dosovitskiy A, Ilg E, Häusser P, Hazirbas C, Golkov V, van der Smagt P, Cremers D and Brox T 2015 Flownet: learning optical flow with convolutional networks CoRR (arXiv:1504.06852)
[10] He K, Zhang X, Ren S and Sun J 2015 Deep residual learning for image recognition CoRR (arXiv:1512.03385)
[11] Ilg E, Mayer N, Saikia T, Keuper M, Dosovitskiy A and Brox T 2016 Flownet 2.0: evolution of optical flow estimation with deep networks CoRR (arXiv:1612.01925)
[12] Jaderberg M, Simonyan K, Zisserman A and Kavukcuoglu K 2015 Spatial transformer networks CoRR (arXiv:1506.02025)
[13] Kalchbrenner N, van den Oord A, Simonyan K, Danihelka I, Vinyals O, Graves A and Kavukcuoglu K 2016 Video pixel networks CoRR (arXiv:1610.00527)
[14] Karl M, Soelch M, Bayer J and van der Smagt P 2016 Deep variational bayes filters: unsupervised learning of state space models from raw data ICLR 2017
[15] Kingma D P and Welling M 2014 Auto-encoding variational bayes ICLR 2014
[16] Krishnan R G, Shalit U and Sontag D 2015 Deep kalman filters (arXiv:1511.05121)
[17] Ladický L, Jeong S H, Solenthaler B, Pollefeys M and Gross M 2015 Data-driven fluid simulations using regression forests ACM Trans. Graph.34 199:1-199:9 · doi:10.1145/2816795.2818129
[18] Ling H and Okada K 2006 Diffusion distance for histogram comparison IEEE Computer Society Conf. on Computer Vision and Pattern Recognitionvol 1 pp 246-53
[19] Madec G 2008 NEMO Ocean Engine(Note du Pôle de Modélisation vol 27) (France: Institut Pierre-Simon Laplace)
[20] Mathieu M, Couprie C and LeCun Y 2015 Deep multi-scale video prediction beyond mean square error CoRR (arXiv:1511.05440)
[21] Patraucean V, Handa A and Cipolla R 2015 Spatio-temporal video autoencoder with differentiable memory CoRR (arXiv:1511.06309)
[22] Raissi M, Perdikaris P and Karniadakis G E 2017 Numerical gaussian processes for time-dependent and non-linear partial differential equations (arXiv:1703.10230)
[23] Rudy S H, Brunton S L, Proctor J L and Nathan Kutz J 2017 Data-driven discovery of partial differential equations Sci. Adv.3 e1602614 · doi:10.1126/sciadv.1602614
[24] Shi X, Chen Z, Wang H, Yeung D Y, Wong W K and Woo W C 2015 Convolutional LSTM network: a machine learning approach for precipitation nowcasting CoRR (arXiv:1506.04214)
[25] Sun D, Roth S, Lewis J P and Black M J 2008 Learning Optical Flow (Berlin: Springer) pp 83-97
[26] Tompson J, Schlachter K, Sprechmann P and Perlin K 2017 Accelerating Eulerian fluid simulation with convolutional networks Proc. of the 34th Int. Conf. on Machine Learning(Australia, 6-11 August 2017) ed D Precup and Y Whye (Sydney: International Convention Centre) pp 3424-33
[27] Trémolet Y 2006 Accounting for an imperfect model in 4D-Var Q. J. R. Meteorol. Soc.132 2483-504 · doi:10.1256/qj.05.224
[28] Vallis G K 2017 Atmospheric and Oceanic Fluid Dynamics (Cambridge: Cambridge University Press) · Zbl 1374.86002 · doi:10.1017/9781107588417
[29] van Amersfoort J, Kannan A, Ranzato M A, Szlam A, Tran D and Chintala S 2017 Transformation-based models of video sequences CoRR pp 1-11 (arXiv:1701.08435)
[30] Walker J, Doersch C, Gupta A and Hebert M 2016 An uncertain future: forecasting from static images using variational autoencoders European Conf. on Computer Vision · doi:10.1007/978-3-319-46478-7_51
[31] Yu J J, Harley A W and Derpanis K G 2016 Back to Basics: Unsupervised Learning of Optical Flow via Brightness Constancy and Motion Smoothness (Cham: Springer) pp 3-10
[32] Zhu Y, Lan Z Z, Newsam S D and Hauptmann A G 2017 Guided optical flow learning CoRR (arXiv:1702.02295)
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.