{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,6]],"date-time":"2026-05-06T02:36:43Z","timestamp":1778035003258,"version":"3.51.4"},"reference-count":46,"publisher":"Springer Science and Business Media LLC","issue":"2","license":[{"start":{"date-parts":[[2025,1,30]],"date-time":"2025-01-30T00:00:00Z","timestamp":1738195200000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2025,1,30]],"date-time":"2025-01-30T00:00:00Z","timestamp":1738195200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"name":"Dutch Foundation of Science","award":["VI.C. 202-031"],"award-info":[{"award-number":["VI.C. 202-031"]}]},{"name":"Dutch Foundation of Science","award":["VI.C. 202-031"],"award-info":[{"award-number":["VI.C. 202-031"]}]}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["J Math Imaging Vis"],"published-print":{"date-parts":[[2025,4]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>PDE-based group convolutional neural networks (PDE-G-CNNs) use solvers of evolution PDEs as substitutes for the conventional components in G-CNNs. PDE-G-CNNs can offer several benefits simultaneously: fewer parameters, inherent equivariance, better accuracy, and data efficiency. In this article, we focus on Euclidean equivariant PDE-G-CNNs where the feature maps are two-dimensional throughout. We call this variant of the framework a PDE-CNN. From a machine learning perspective, we list several practically desirable axioms and derive from these which PDEs should be used in a PDE-CNN, this being our main contribution. Our approach to geometric learning via PDEs is inspired by the axioms of scale-space theory, which we generalize by introducing semifield-valued signals. Our theory reveals new PDEs that can be used in PDE-CNNs and we experimentally examine what impact these have on the accuracy of PDE-CNNs. We also confirm for small networks that PDE-CNNs offer fewer parameters, increased accuracy, and better data efficiency when compared to CNNs.<\/jats:p>","DOI":"10.1007\/s10851-025-01230-4","type":"journal-article","created":{"date-parts":[[2025,1,30]],"date-time":"2025-01-30T02:54:28Z","timestamp":1738205668000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":3,"title":["PDE-CNNs: Axiomatic Derivations and Applications"],"prefix":"10.1007","volume":"67","author":[{"given":"Gijs","family":"Bellaard","sequence":"first","affiliation":[]},{"given":"Sei","family":"Sakata","sequence":"additional","affiliation":[]},{"given":"Bart M. N.","family":"Smets","sequence":"additional","affiliation":[]},{"given":"Remco","family":"Duits","sequence":"additional","affiliation":[]}],"member":"297","published-online":{"date-parts":[[2025,1,30]]},"reference":[{"key":"1230_CR1","doi-asserted-by":"publisher","first-page":"209","DOI":"10.1007\/s10851-022-01114-x","volume":"65","author":"B Smets","year":"2023","unstructured":"Smets, B., Portegies, J., Bekkers, E., Duits, R.: PDE-based group equivariant convolutional neural networks. J. Math. Imaging Vis. 65, 209\u2013239 (2023). https:\/\/doi.org\/10.1007\/s10851-022-01114-x","journal-title":"J. Math. Imaging Vis."},{"key":"1230_CR2","unstructured":"Cohen, T., Welling, M.: Group equivariant convolutional networks. (eds Balcan, M.F., Weinberger, K.Q.) Proceedings of The 33rd International Conference on Machine Learning, Vol. 48 of Proceedings of Machine Learning Research, 2990\u20132999 (PMLR, New York, New York, USA, 2016)"},{"key":"1230_CR3","doi-asserted-by":"publisher","unstructured":"Pai, G., Bellaard, G., Smets, B., Duits, R.: Functional properties of PDE-based group equivariant convolutional neural networks. (eds Nielsen, F. & Barbaresco, F.) Geometric Science of Information, 63\u201372 (Springer Nature Switzerland, Cham, 2023). https:\/\/doi.org\/10.1007\/978-3-031-38271-0_7","DOI":"10.1007\/978-3-031-38271-0_7"},{"key":"1230_CR4","doi-asserted-by":"publisher","first-page":"819","DOI":"10.1007\/s10851-023-01147-w","volume":"65","author":"G Bellaard","year":"2023","unstructured":"Bellaard, G., Bon, D., Pai, G., Smets, B., Duits, R.: Analysis of (sub-)Riemannian PDE-G-CNNs. J. Math. Imaging Vis. 65, 819\u2013843 (2023). https:\/\/doi.org\/10.1007\/s10851-023-01147-w","journal-title":"J. Math. Imaging Vis."},{"key":"1230_CR5","doi-asserted-by":"publisher","unstructured":"Bellaard, G., Pai, G., Bescos, J., Duits, R.: Geometric adaptations of PDE-G-CNNs. (eds Calatroni, L., Donatelli, M., Morigi, S., Prato, M. & Santacesaria, M.) Scale Space and Variational Methods in Computer Vision, 538\u2013550 (Springer International Publishing, Cham, 2023). https:\/\/doi.org\/10.1007\/978-3-031-31975-4_41","DOI":"10.1007\/978-3-031-31975-4_41"},{"key":"1230_CR6","doi-asserted-by":"publisher","unstructured":"Mohamed, M., Cesa, G., Cohen, T., Welling, M.: A data and compute efficient design for limited-resources deep learning (2020). https:\/\/doi.org\/10.48550\/arXiv.2004.09691","DOI":"10.48550\/arXiv.2004.09691"},{"key":"1230_CR7","unstructured":"Cohen, T., Weiler, M., Kicanaoglu, B., Welling, M.: Gauge equivariant convolutional networks and the icosahedral CNN. (eds Chaudhuri, K. & Salakhutdinov, R.) In Proceedings of the 36th International Conference on Machine Learning, Vol.\u00a097, 1321\u20131330 (2019). https:\/\/proceedings.mlr.press\/v97\/cohen19d.html"},{"key":"1230_CR8","doi-asserted-by":"publisher","first-page":"691","DOI":"10.1109\/34.391411","volume":"17","author":"E Pauwels","year":"1995","unstructured":"Pauwels, E., van Gool, L., Fiddelaers, P., Moons, T.: An extended class of scale-invariant and recursive scale space filters. IEEE Trans. Pattern Anal. Mach. Intell. 17, 691\u2013701 (1995). https:\/\/doi.org\/10.1109\/34.391411","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"1230_CR9","unstructured":"Iijima, T.: Basic theory of pattern observation. Papers of Technical Group on Automata and Automatic Control (1959)"},{"key":"1230_CR10","doi-asserted-by":"publisher","first-page":"199","DOI":"10.1007\/BF00375127","volume":"123","author":"L Alvarez","year":"1993","unstructured":"Alvarez, L., Guichard, F., Lions, P., Morel, J.: Axioms and fundamental equations of image processing. Arch. Ration. Mech. Anal. 123, 199\u2013257 (1993). https:\/\/doi.org\/10.1007\/BF00375127","journal-title":"Arch. Ration. Mech. Anal."},{"key":"1230_CR11","doi-asserted-by":"publisher","first-page":"39","DOI":"10.1023\/A:1011185417206","volume":"42","author":"L Florack","year":"2001","unstructured":"Florack, L.: Non-linear scale-spaces isomorphic to the linear case with applications to scalar, vector and multispectral images. Int. J. Comput. Vision 42, 39\u201353 (2001). https:\/\/doi.org\/10.1023\/A:1011185417206","journal-title":"Int. J. Comput. Vision"},{"key":"1230_CR12","doi-asserted-by":"publisher","first-page":"269","DOI":"10.1006\/jvci.2001.0480","volume":"13","author":"H Heijmans","year":"2002","unstructured":"Heijmans, H., Boomgaard, R.: Algebraic framework for linear and morphological scale-spaces. J. Vis. Commun. Image Represent. 13, 269\u2013301 (2002). https:\/\/doi.org\/10.1006\/jvci.2001.0480","journal-title":"J. Vis. Commun. Image Represent."},{"key":"1230_CR13","doi-asserted-by":"publisher","first-page":"267","DOI":"10.1023\/B:JMIV.0000024043.96722.aa","volume":"20","author":"R Duits","year":"2004","unstructured":"Duits, R., Florack, L., de Graaf, J., Haar Romenij, B.: On the axioms of scale space theory. J. Math. Imaging Vis. 20, 267\u2013298 (2004). https:\/\/doi.org\/10.1023\/B:JMIV.0000024043.96722.aa","journal-title":"J. Math. Imaging Vis."},{"key":"1230_CR14","doi-asserted-by":"publisher","first-page":"5","DOI":"10.1023\/B:JMIV.0000026554.79537.35","volume":"21","author":"M Felsberg","year":"2004","unstructured":"Felsberg, M., Sommer, G.: The monogenic scale-space: a unifying approach to phase-based image processing in scale-space. J. Math. Imaging Vis. 21, 5\u201326 (2004). https:\/\/doi.org\/10.1023\/B:JMIV.0000026554.79537.35","journal-title":"J. Math. Imaging Vis."},{"key":"1230_CR15","doi-asserted-by":"publisher","first-page":"237","DOI":"10.1023\/A:1008344623873","volume":"10","author":"J Weickert","year":"1999","unstructured":"Weickert, J., Ishikawa, S., Imiya, A.: Linear scale-space has first been proposed in japan. J. Math. Imaging Vis. 10, 237\u2013252 (1999). https:\/\/doi.org\/10.1023\/A:1008344623873","journal-title":"J. Math. Imaging Vis."},{"key":"1230_CR16","doi-asserted-by":"publisher","first-page":"363","DOI":"10.1007\/BF00336961","volume":"50","author":"J Koenderink","year":"1984","unstructured":"Koenderink, J.: The structure of images. Biol. Cybern. 50, 363\u2013370 (1984). https:\/\/doi.org\/10.1007\/BF00336961","journal-title":"Biol. Cybern."},{"key":"1230_CR17","doi-asserted-by":"publisher","unstructured":"Brockett, R., Maragos, P.: Evolution equations for continuous-scale morphology. (ed.IEEE) In 1992 IEEE International Conference on Acoustics, Speech, and Signal Processing, Vol.\u00a03, 125\u2013128 vol. 3 (1992). https:\/\/doi.org\/10.1109\/ICASSP.1992.226260","DOI":"10.1109\/ICASSP.1992.226260"},{"key":"1230_CR18","doi-asserted-by":"publisher","first-page":"501","DOI":"10.1109\/TMI.2004.825627","volume":"23","author":"J Staal","year":"2004","unstructured":"Staal, J., Abramoff, M., Niemeijer, M., Viergever, M., van Ginneken, B.: Ridge-based vessel segmentation in color images of the retina. IEEE Trans. Med. Imaging 23, 501\u2013509 (2004). https:\/\/doi.org\/10.1109\/TMI.2004.825627","journal-title":"IEEE Trans. Med. Imaging"},{"key":"1230_CR19","doi-asserted-by":"publisher","unstructured":"Welk, M.: Families of generalised morphological scale spaces. (eds Griffin, L. & Lillholm, M.) Scale Space Methods in Computer Vision, 770\u2013784 (Springer, Berlin, Heidelberg, 2003). https:\/\/doi.org\/10.1007\/3-540-44935-3_54","DOI":"10.1007\/3-540-44935-3_54"},{"key":"1230_CR20","doi-asserted-by":"publisher","first-page":"352","DOI":"10.1007\/s10851-016-0646-8","volume":"56","author":"M Schmidt","year":"2016","unstructured":"Schmidt, M., Weickert, J.: Morphological counterparts of linear shift-invariant scale-spaces. J. Mat. Imaging Vis. 56, 352\u2013366 (2016). https:\/\/doi.org\/10.1007\/s10851-016-0646-8","journal-title":"J. Mat. Imaging Vis."},{"key":"1230_CR21","doi-asserted-by":"publisher","first-page":"79","DOI":"10.1016\/0165-1684(94)90058-2","volume":"38","author":"L Dorst","year":"1994","unstructured":"Dorst, L., Boomgaard, R.: Morphological signal processing and the slope transform. Signal Process. 38, 79\u201398 (1994). https:\/\/doi.org\/10.1016\/0165-1684(94)90058-2","journal-title":"Signal Process."},{"key":"1230_CR22","doi-asserted-by":"publisher","first-page":"157","DOI":"10.1007\/s11263-005-1841-z","volume":"64","author":"B Burgeth","year":"2005","unstructured":"Burgeth, B., Weickert, J.: An explanation for the logarithmic connection between linear and morphological system theory. Int. J. Comput. Vision 64, 157\u2013169 (2005). https:\/\/doi.org\/10.1007\/s11263-005-1841-z","journal-title":"Int. J. Comput. Vision"},{"key":"1230_CR23","doi-asserted-by":"publisher","unstructured":"Sapiro, G.: Geometric Partial Differential Equations and Image Analysis (Cambridge University Press, 2001). https:\/\/doi.org\/10.1017\/CBO9780511626319","DOI":"10.1017\/CBO9780511626319"},{"key":"1230_CR24","doi-asserted-by":"publisher","first-page":"212","DOI":"10.1137\/15M1013572","volume":"9","author":"G Citti","year":"2016","unstructured":"Citti, G., Franceschiello, B., Sanguinetti, G., Sarti, A.: Sub-riemannian mean curvature flow for image processing. SIAM J. Imag. Sci. 9, 212\u2013237 (2016). https:\/\/doi.org\/10.1137\/15M1013572","journal-title":"SIAM J. Imag. Sci."},{"key":"1230_CR25","doi-asserted-by":"publisher","unstructured":"Guichard, F., Morel, J.: Partial differential equations and image iterative filtering. (eds Duff, I. & Watson, G.) The State of the Art in Numerical Analysis., Vol.\u00a063, 252\u2013562 (Clarendon Press, 1997). https:\/\/doi.org\/10.1093\/oso\/9780198500148.003.0020","DOI":"10.1093\/oso\/9780198500148.003.0020"},{"key":"1230_CR26","doi-asserted-by":"publisher","unstructured":"Evans, L.: Partial Differential Equations Graduate studies in mathematics (American Mathematical Society, 2010). https:\/\/doi.org\/10.1090\/gsm\/019","DOI":"10.1090\/gsm\/019"},{"key":"1230_CR27","doi-asserted-by":"publisher","unstructured":"Jacobsen, J.-H., Van\u00a0Gemert, J., Lou, Z., Smeulders, A.W.M.: Structured receptive fields in cnns. (ed.IEEE) In 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Conference on Computer Vision and Pattern Recognition, 2610\u20132619 (IEEE, 2016). https:\/\/doi.org\/10.1109\/CVPR.2016.286","DOI":"10.1109\/CVPR.2016.286"},{"key":"1230_CR28","doi-asserted-by":"publisher","first-page":"8342","DOI":"10.1109\/TIP.2021.3115001","volume":"30","author":"SL Pintea","year":"2021","unstructured":"Pintea, S.L., Tomen, N., Goes, S.F., Loog, M., van Gemert, J.C.: Resolution learning in deep convolutional networks using Scale-Space theory. IEEE Trans. Image Process. 30, 8342\u20138353 (2021). https:\/\/doi.org\/10.1109\/TIP.2021.3115001","journal-title":"IEEE Trans. Image Process."},{"key":"1230_CR29","unstructured":"Tomen, N., Pintea, S.-L., Van\u00a0Gemert, J.: Deep continuous networks. (eds Meila, M. & Zhang, T.) In Proceedings of the 38th International Conference on Machine Learning, Vol. 139 of Proceedings of Machine Learning Research, 10324\u201310335 (2021). https:\/\/proceedings.mlr.press\/v139\/tomen21a.html"},{"key":"1230_CR30","unstructured":"Saldanha, N., Pintea, S.L., van Gemert, J.C., Tomen, N.: Frequency learning for structured cnn filters with gaussian fractional derivatives (2021). https:\/\/arxiv.org\/abs\/2111.06660. arXiv:2111.06660"},{"key":"1230_CR31","doi-asserted-by":"crossref","unstructured":"Basting, M., et\u00a0al.: Scale Learning in Scale-Equivariant Convolutional Networks. Master\u2019s thesis, Delft University of Technology (2023)","DOI":"10.5220\/0012379800003660"},{"key":"1230_CR32","doi-asserted-by":"publisher","first-page":"4357","DOI":"10.1109\/TIP.2018.2835143","volume":"27","author":"S Luan","year":"2018","unstructured":"Luan, S., Chen, C., Zhang, B., Han, J., Liu, J.: Gabor convolutional networks. IEEE Trans. Image Process. 27, 4357\u20134366 (2018). https:\/\/doi.org\/10.1109\/TIP.2018.2835143","journal-title":"IEEE Trans. Image Process."},{"key":"1230_CR33","unstructured":"Romero, D.W., et\u00a0al.: Flexconv: Continuous kernel convolutions with differentiable kernel sizes. (ed.ICLR) In 10th International Conference on Learning Representations (ICLR) (2022). https:\/\/openreview.net\/forum?id=3jooF27-0Wy"},{"key":"1230_CR34","unstructured":"Worrall, D., Welling, M.: Deep scale-spaces: Equivariance over scale. (eds Wallach, H. et\u00a0al.) Advances in Neural Information Processing Systems, Vol.\u00a032 (Curran Associates, Inc.,). (2019) https:\/\/proceedings.neurips.cc\/paper_files\/paper\/2019\/file\/f04cd7399b2b0128970efb6d20b5c551-Paper.pdf"},{"key":"1230_CR35","unstructured":"Sosnovik, I., Szmaja, M., Smeulders, A.: Scale-equivariant steerable networks. (ed.ICLR) In International Conference on Learning Representations (2020). https:\/\/openreview.net\/forum?id=HJgpugrKPS"},{"key":"1230_CR36","doi-asserted-by":"publisher","first-page":"223","DOI":"10.1007\/s10851-021-01057-9","volume":"64","author":"T Lindeberg","year":"2022","unstructured":"Lindeberg, T.: Scale-covariant and scale-invariant gaussian derivative networks. J. Math. Imaging Vis. 64, 223\u2013242 (2022). https:\/\/doi.org\/10.1007\/s10851-021-01057-9","journal-title":"J. Math. Imaging Vis."},{"key":"1230_CR37","doi-asserted-by":"publisher","unstructured":"Sangalli, M., Blusseau, S., Velasco-Forero, S., Angulo, J.: Scale equivariant neural networks with morphological scale-spaces. (eds Lindblad, J., Malmberg, F. & Sladoje, N.) Discrete Geometry and Mathematical Morphology, 483\u2013495 (Springer International Publishing, Cham, 2021). https:\/\/doi.org\/10.1007\/978-3-030-76657-3_35","DOI":"10.1007\/978-3-030-76657-3_35"},{"key":"1230_CR38","unstructured":"saz. Measurability of a pointwise limit of measurable functions. Mathematics Stack Exchange. https:\/\/math.stackexchange.com\/q\/921576"},{"key":"1230_CR39","unstructured":"Matematleta. a proof that the pointswise limits of lower semicontinuous (lsc) functions is lsc. Mathematics Stack Exchange. https:\/\/math.stackexchange.com\/q\/2968282"},{"key":"1230_CR40","doi-asserted-by":"crossref","unstructured":"Lindeberg, T.: On the Axiomatic Foundations of Linear Scale-Space, 75\u201397 (Springer Netherlands, Dordrecht, 1997). https:\/\/doi.org\/10.1007\/978-94-015-8802-7_6","DOI":"10.1007\/978-94-015-8802-7_6"},{"key":"1230_CR41","doi-asserted-by":"publisher","unstructured":"Kolokoltsov, V., Maslov, V.: Idempotent Analysis and Its Applications (Springer. Dordrecht, 1997). https:\/\/doi.org\/10.1007\/978-94-015-8901-7","DOI":"10.1007\/978-94-015-8901-7"},{"key":"1230_CR42","unstructured":"Rudin, W.: Functional Analysis (McGraw-Hill, 1973, 1991)"},{"key":"1230_CR43","unstructured":"Kraakman, T.B.: On the Construction of Classical and Quantum PDE-Based Neural Networks. Master\u2019s thesis, Eindhoven University of Technology (2023)"},{"key":"1230_CR44","doi-asserted-by":"publisher","unstructured":"Lindeberg, T., G\u00e5rding, J.: Shape-adapted smoothing in estimation of 3-d shape cues from affine deformations of local 2-d brightness structure. Image and Vision Computing15, 415\u2013434 (1997). https:\/\/www.sciencedirect.com\/science\/article\/pii\/S026288569701144X. https:\/\/doi.org\/10.1016\/S0262-8856(97)01144-X","DOI":"10.1016\/S0262-8856(97)01144-X"},{"key":"1230_CR45","unstructured":"Castella, A.: Introduction to Shift-Invariant PDE-based Convolutional Neural Networks and experimentation with Vessel Segmentation. Bachelor\u2019s thesis, Eindhoven University of Technology (2021)"},{"key":"1230_CR46","doi-asserted-by":"publisher","unstructured":"Borwein, J., Lewis, A.: Convex Analysis and Nonlinear Optimization (Springer New York, NY, 2016). https:\/\/doi.org\/10.1007\/978-0-387-31256-9","DOI":"10.1007\/978-0-387-31256-9"}],"container-title":["Journal of Mathematical Imaging and Vision"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s10851-025-01230-4.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/article\/10.1007\/s10851-025-01230-4\/fulltext.html","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1007\/s10851-025-01230-4.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,4,23]],"date-time":"2025-04-23T03:41:29Z","timestamp":1745379689000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1007\/s10851-025-01230-4"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,1,30]]},"references-count":46,"journal-issue":{"issue":"2","published-print":{"date-parts":[[2025,4]]}},"alternative-id":["1230"],"URL":"https:\/\/doi.org\/10.1007\/s10851-025-01230-4","relation":{"has-preprint":[{"id-type":"doi","id":"10.21203\/rs.3.rs-4269855\/v1","asserted-by":"object"}]},"ISSN":["0924-9907","1573-7683"],"issn-type":[{"value":"0924-9907","type":"print"},{"value":"1573-7683","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,1,30]]},"assertion":[{"value":"15 April 2024","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"15 January 2025","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"30 January 2025","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"R. Duits is a member of the editorial board of Journal of Mathematical Imaging and Vision (JMIV).","order":2,"name":"Ethics","group":{"name":"EthicsHeading","label":"Conflict of interest"}}],"article-number":"13"}}