{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,25]],"date-time":"2026-05-25T07:04:57Z","timestamp":1779692697387,"version":"3.53.1"},"reference-count":54,"publisher":"Elsevier BV","license":[{"start":{"date-parts":[[2026,8,1]],"date-time":"2026-08-01T00:00:00Z","timestamp":1785542400000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/www.elsevier.com\/tdm\/userlicense\/1.0\/"},{"start":{"date-parts":[[2026,8,1]],"date-time":"2026-08-01T00:00:00Z","timestamp":1785542400000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/www.elsevier.com\/legal\/tdmrep-license"},{"start":{"date-parts":[[2026,8,1]],"date-time":"2026-08-01T00:00:00Z","timestamp":1785542400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-017"},{"start":{"date-parts":[[2026,8,1]],"date-time":"2026-08-01T00:00:00Z","timestamp":1785542400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"},{"start":{"date-parts":[[2026,8,1]],"date-time":"2026-08-01T00:00:00Z","timestamp":1785542400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-012"},{"start":{"date-parts":[[2026,8,1]],"date-time":"2026-08-01T00:00:00Z","timestamp":1785542400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,8,1]],"date-time":"2026-08-01T00:00:00Z","timestamp":1785542400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-004"}],"funder":[{"DOI":"10.13039\/501100002241","name":"Japan Science and Technology Agency","doi-asserted-by":"publisher","award":["JPMJAP2329"],"award-info":[{"award-number":["JPMJAP2329"]}],"id":[{"id":"10.13039\/501100002241","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100002241","name":"Japan Science and Technology Agency","doi-asserted-by":"publisher","award":["CREST JPMJCR24Q5"],"award-info":[{"award-number":["CREST JPMJCR24Q5"]}],"id":[{"id":"10.13039\/501100002241","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100000038","name":"Natural Sciences and Engineering Research Council of Canada","doi-asserted-by":"publisher","award":["RGPIN-2023-04482"],"award-info":[{"award-number":["RGPIN-2023-04482"]}],"id":[{"id":"10.13039\/501100000038","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100000038","name":"Natural Sciences and Engineering Research Council of Canada","doi-asserted-by":"publisher","award":["DGECR-2023-00230"],"award-info":[{"award-number":["DGECR-2023-00230"]}],"id":[{"id":"10.13039\/501100000038","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100019117","name":"Vector Institute","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100019117","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100007631","name":"Canadian Institute for Advanced Research","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100007631","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001691","name":"Japan Society for the Promotion of Science","doi-asserted-by":"publisher","award":["25H01453"],"award-info":[{"award-number":["25H01453"]}],"id":[{"id":"10.13039\/501100001691","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001691","name":"Japan Society for the Promotion of Science","doi-asserted-by":"publisher","award":["JP24K16949"],"award-info":[{"award-number":["JP24K16949"]}],"id":[{"id":"10.13039\/501100001691","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["elsevier.com","sciencedirect.com"],"crossmark-restriction":true},"short-container-title":["Neural Networks"],"published-print":{"date-parts":[[2026,8]]},"DOI":"10.1016\/j.neunet.2026.108797","type":"journal-article","created":{"date-parts":[[2026,3,5]],"date-time":"2026-03-05T17:12:04Z","timestamp":1772730724000},"page":"108797","update-policy":"https:\/\/doi.org\/10.1016\/elsevier_cm_policy","source":"Crossref","is-referenced-by-count":0,"special_numbering":"C","title":["Approximation rates in Besov norms and sample-complexity of Kolmogorov-Arnold networks with residual connections"],"prefix":"10.1016","volume":"200","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-6791-3371","authenticated-orcid":false,"given":"Anastasis","family":"Kratsios","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4155-9225","authenticated-orcid":false,"given":"Bum Jun","family":"Kim","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6132-6846","authenticated-orcid":false,"given":"Takashi","family":"Furuya","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"78","reference":[{"issue":"2","key":"10.1016\/j.neunet.2026.108797_bib0001","doi-asserted-by":"crossref","first-page":"671","DOI":"10.1111\/mafi.12389","article-title":"Designing universal causal deep learning models: The geometric (hyper) transformer","volume":"34","author":"Acciaio","year":"2024","journal-title":"Mathematical Finance"},{"issue":"4","key":"10.1016\/j.neunet.2026.108797_bib0002","doi-asserted-by":"crossref","first-page":"615","DOI":"10.1145\/263867.263927","article-title":"Scale-sensitive dimensions, uniform convergence, and learnability","volume":"44","author":"Alon","year":"1997","journal-title":"Journal of the ACM"},{"key":"10.1016\/j.neunet.2026.108797_bib0003","series-title":"Neural network learning: Theoretical foundations","author":"Anthony","year":"2009"},{"key":"10.1016\/j.neunet.2026.108797_bib0004","first-page":"190","article-title":"Almost linear VC dimension bounds for piecewise polynomial networks","volume":"11","author":"Bartlett","year":"1998","journal-title":"Advances in Neural Information Processing Systems"},{"issue":"5","key":"10.1016\/j.neunet.2026.108797_bib0005","doi-asserted-by":"crossref","first-page":"1721","DOI":"10.1109\/18.623181","article-title":"Covering numbers for real-valued function classes","volume":"43","author":"Bartlett","year":"1997","journal-title":"IEEE Transactions on Information Theory"},{"key":"10.1016\/j.neunet.2026.108797_bib0006","doi-asserted-by":"crossref","first-page":"242","DOI":"10.1016\/j.neunet.2023.01.035","article-title":"Simultaneous approximation of a smooth function and its derivatives by deep neural networks with piecewise-polynomial activations","volume":"161","author":"Belomestny","year":"2023","journal-title":"Neural Networks"},{"issue":"7","key":"10.1016\/j.neunet.2026.108797_bib0007","doi-asserted-by":"crossref","first-page":"4314","DOI":"10.1016\/j.jfa.2014.02.001","article-title":"Extending sobolev functions with partially vanishing traces from locally (\u03b5, \u03b4)-domains and applications to mixed boundary problems","volume":"266","author":"Brewster","year":"2014","journal-title":"Journal of Functional Analysis"},{"issue":"7","key":"10.1016\/j.neunet.2026.108797_bib0008","doi-asserted-by":"crossref","first-page":"3079","DOI":"10.1109\/TNNLS.2021.3049719","article-title":"Efficient approximation of high-dimensional functions with neural networks","volume":"33","author":"Cheridito","year":"2021","journal-title":"IEEE Transactions on Neural Networks and Learning Systems"},{"issue":"3","key":"10.1016\/j.neunet.2026.108797_bib0009","doi-asserted-by":"crossref","first-page":"263","DOI":"10.1016\/0021-9045(92)90120-D","article-title":"A general framework of compactly supported splines and wavelets","volume":"71","author":"Chui","year":"1992","journal-title":"Journal of Approximation Theory"},{"key":"10.1016\/j.neunet.2026.108797_bib0010","article-title":"A practical guide to splines","volume":"vol. 27","author":"de Boor","year":"2001"},{"key":"10.1016\/j.neunet.2026.108797_bib0011","unstructured":"de Haitz S\u00e1ez, B., Lukoianov, A., Kratsios, A., Bronstein, M., & Dong, X. (2024). Scalable message passing neural networks: No need for attention in large graph representation learning. arXiv preprint arXiv: 2411.00835."},{"key":"10.1016\/j.neunet.2026.108797_bib0012","doi-asserted-by":"crossref","first-page":"732","DOI":"10.1016\/j.neunet.2021.08.015","article-title":"On the approximation of functions by tanh neural networks","volume":"143","author":"De Ryck","year":"2021","journal-title":"Neural Networks"},{"key":"10.1016\/j.neunet.2026.108797_bib0013","series-title":"Constructive theory of functions (Varna, 1987)","first-page":"110","article-title":"Interpolation of approximation spaces","author":"DeVore","year":"1988"},{"issue":"2","key":"10.1016\/j.neunet.2026.108797_bib0014","first-page":"843","article-title":"Besov spaces on domains in Rd","volume":"335","author":"DeVore","year":"1993","journal-title":"Transactions of the American Mathematical Society"},{"key":"10.1016\/j.neunet.2026.108797_bib0015","unstructured":"Furuya, T., & Kratsios, A. (2024). Simultaneously solving FBSDEs with neural operators of logarithmic depth, constant width, and sub-linear rank. arXiv preprint arXiv: 2410.14788."},{"issue":"1","key":"10.1016\/j.neunet.2026.108797_bib0016","doi-asserted-by":"crossref","first-page":"259","DOI":"10.1007\/s00365-021-09543-4","article-title":"Approximation spaces of deep neural networks","volume":"55","author":"Gribonval","year":"2022","journal-title":"Constructive Approximation"},{"key":"10.1016\/j.neunet.2026.108797_bib0017","series-title":"Proceedings of the IEEE conference on computer vision and pattern recognition","first-page":"770","article-title":"Deep residual learning for image recognition","author":"He","year":"2016"},{"key":"10.1016\/j.neunet.2026.108797_bib0018","unstructured":"Hwang, G. (2023). Minimum width for deep, narrow MLP: A diffeomorphism approach. arXiv preprint arXiv: 2308.15873."},{"issue":"11-12","key":"10.1016\/j.neunet.2026.108797_bib0019","doi-asserted-by":"crossref","first-page":"1303","DOI":"10.1002\/mana.201100245","article-title":"Smoothness spaces of higher order on lower dimensional subsets of the Euclidean space","volume":"288","author":"Ihnatsyeva","year":"2015","journal-title":"Mathematische Nachrichten"},{"issue":"9","key":"10.1016\/j.neunet.2026.108797_bib0020","doi-asserted-by":"crossref","first-page":"1870","DOI":"10.1016\/j.jfa.2013.07.006","article-title":"Characterization of traces of smooth functions on Ahlfors regular sets","volume":"265","author":"Ihnatsyeva","year":"2013","journal-title":"Journal of Functional Analysis"},{"issue":"4","key":"10.1016\/j.neunet.2026.108797_bib0021","doi-asserted-by":"crossref","first-page":"3635","DOI":"10.1137\/21M144431X","article-title":"Deep neural networks with relu-sine-exponential activations break curse of dimensionality in approximation on h\u00f6lder class","volume":"55","author":"Jiao","year":"2023","journal-title":"SIAM Journal on Mathematical Analysis"},{"key":"10.1016\/j.neunet.2026.108797_bib0022","doi-asserted-by":"crossref","first-page":"71","DOI":"10.1007\/BF02392869","article-title":"Quasiconformal mappings and extendability of functions in Sobolev spaces","volume":"147","author":"Jones","year":"1981","journal-title":"Acta Mathematica"},{"key":"10.1016\/j.neunet.2026.108797_bib0023","doi-asserted-by":"crossref","first-page":"229","DOI":"10.1016\/0021-9045(75)90035-0","article-title":"Sur le th\u00e9or\u00e8me de superposition de Kolmogorov","volume":"13","author":"Kahane","year":"1975","journal-title":"Journal of Approximation Theory"},{"issue":"1","key":"10.1016\/j.neunet.2026.108797_bib0024","doi-asserted-by":"crossref","first-page":"169","DOI":"10.1006\/jcss.1997.1477","article-title":"Polynomial bounds for VC dimension of sigmoidal and general pfaffian neural networks","volume":"54","author":"Karpinski","year":"1997","journal-title":"Journal of Computer and System Sciences"},{"key":"10.1016\/j.neunet.2026.108797_bib0025","series-title":"Conference on learning theory","first-page":"2306","article-title":"Universal approximation with deep narrow networks","author":"Kidger","year":"2020"},{"key":"10.1016\/j.neunet.2026.108797_bib0026","series-title":"ICML 2024 workshop on in-context learning","article-title":"Transformers are minimax optimal nonparametric in-context learners","author":"Kim","year":"2024"},{"issue":"1","key":"10.1016\/j.neunet.2026.108797_bib0027","first-page":"8896","article-title":"Universal approximation theorems for differentiable geometric deep learning","volume":"23","author":"Kratsios","year":"2022","journal-title":"The Journal of Machine Learning Research"},{"key":"10.1016\/j.neunet.2026.108797_bib0028","first-page":"6172","article-title":"Resnet with one-neuron hidden layers is a universal approximator","volume":"31","author":"Lin","year":"2018","journal-title":"Advances in Neural Information Processing Systems"},{"key":"10.1016\/j.neunet.2026.108797_bib0029","series-title":"The thirteenth international conference on learning representations","article-title":"KAN: Kolmogorov\u2013arnold networks","author":"Liu","year":"2025"},{"issue":"1","key":"10.1016\/j.neunet.2026.108797_bib0030","first-page":"69","article-title":"Multiresolution approximations and wavelet orthonormal bases of l2","volume":"315","author":"Mallat","year":"1989","journal-title":"Transactions of the American Mathematical Society"},{"key":"10.1016\/j.neunet.2026.108797_bib0031","unstructured":"Mao, T., Siegel, J. W., & Xu, J. (2024). Approximation rates for shallow ReLUk neural networks on sobolev spaces via the radon transform. arXiv preprint arXiv: 2408.10996."},{"issue":"3","key":"10.1016\/j.neunet.2026.108797_bib0032","doi-asserted-by":"crossref","first-page":"350","DOI":"10.1016\/0196-8858(92)90016-P","article-title":"Approximation by superposition of sigmoidal and radial basis functions","volume":"13","author":"Mhaskar","year":"1992","journal-title":"Advances in Applied Mathematics"},{"key":"10.1016\/j.neunet.2026.108797_bib0033","doi-asserted-by":"crossref","first-page":"229","DOI":"10.1016\/j.neunet.2019.08.028","article-title":"An analysis of training and generalization errors in shallow and deep networks","volume":"121","author":"Mhaskar","year":"2020","journal-title":"Neural Networks"},{"key":"10.1016\/j.neunet.2026.108797_bib0034","series-title":"International conference on learning representations","article-title":"Minimum width for universal approximation","author":"Park","year":"2021"},{"issue":"2","key":"10.1016\/j.neunet.2026.108797_bib0035","doi-asserted-by":"crossref","first-page":"807","DOI":"10.1007\/s00365-022-09601-5","article-title":"Singular value perturbation and deep network optimization","volume":"57","author":"Riedi","year":"2023","journal-title":"Constructive Approximation"},{"issue":"2","key":"10.1016\/j.neunet.2026.108797_bib0036","doi-asserted-by":"crossref","first-page":"619","DOI":"10.1016\/j.jfa.2005.11.013","article-title":"Degree-independent Sobolev extension on locally uniform domains","volume":"235","author":"Rogers","year":"2006","journal-title":"Journal of Functional Analysis"},{"issue":"1","key":"10.1016\/j.neunet.2026.108797_bib0037","doi-asserted-by":"crossref","first-page":"4684","DOI":"10.1038\/s41598-025-88959-5","article-title":"Recurrent Fourier-Kolmogorov Arnold networks for photovoltaic power forecasting","volume":"15","author":"Rong","year":"2025","journal-title":"Scientific Reports"},{"issue":"1","key":"10.1016\/j.neunet.2026.108797_bib0038","doi-asserted-by":"crossref","first-page":"237","DOI":"10.1112\/S0024610799007723","article-title":"On restrictions and extensions of the Besov and Triebel-Lizorkin spaces with respect to Lipschitz domains","volume":"60","author":"Rychkov","year":"1999","journal-title":"Journal of the London Mathematical Society (2)"},{"key":"10.1016\/j.neunet.2026.108797_bib0039","unstructured":"Shazeer, N. (2020). Glu variants improve transformer. arXiv preprint arXiv: 2002.05202."},{"issue":"4","key":"10.1016\/j.neunet.2026.108797_bib0040","doi-asserted-by":"crossref","first-page":"1407","DOI":"10.1002\/mana.202300047","article-title":"New estimates of Rychkov\u2019s universal extension operator for Lipschitz domains and some applications","volume":"297","author":"Shi","year":"2024","journal-title":"Mathematische Nachrichten"},{"key":"10.1016\/j.neunet.2026.108797_bib0041","series-title":"ICLR","article-title":"Adaptivity of deep reLU network for learning in besov and mixed smooth besov spaces: Optimal rate and curse of dimensionality","author":"Suzuki","year":"2019"},{"key":"10.1016\/j.neunet.2026.108797_bib0042","article-title":"Theory of function spaces. III","volume":"vol. 100","author":"Triebel","year":"2006"},{"key":"10.1016\/j.neunet.2026.108797_bib0043","series-title":"Weak convergence","author":"Van Der Vaart","year":"1996"},{"key":"10.1016\/j.neunet.2026.108797_bib0044","first-page":"5998","article-title":"Attention is all you need","volume":"30","author":"Vaswani","year":"2017","journal-title":"Advances in Neural Information Processing Systems"},{"key":"10.1016\/j.neunet.2026.108797_bib0045","article-title":"Don\u2019t fear peculiar activation functions: EUAF and beyond","author":"Wang","year":"2025","journal-title":"Neural Networks"},{"key":"10.1016\/j.neunet.2026.108797_bib0046","series-title":"The thirteenth international conference on learning representations","article-title":"On the expressiveness and spectral bias of KANs","author":"Wang","year":"2025"},{"issue":"1","key":"10.1016\/j.neunet.2026.108797_bib0047","doi-asserted-by":"crossref","first-page":"8648","DOI":"10.1038\/s41598-025-88054-9","article-title":"Graph attention and kolmogorov\u2013arnold network based smart grids intrusion detection","volume":"15","author":"Wu","year":"2025","journal-title":"Scientific Reports"},{"key":"10.1016\/j.neunet.2026.108797_bib0048","first-page":"1","article-title":"Optimal rates of approximation by shallow ReLUk neural networks and applications to nonparametric regression","author":"Yang","year":"2024","journal-title":"Constructive Approximation"},{"key":"10.1016\/j.neunet.2026.108797_bib0049","series-title":"Conference on learning theory","first-page":"639","article-title":"Optimal approximation of continuous functions by very deep ReLU networks","volume":"75","author":"Yarotsky","year":"2018"},{"key":"10.1016\/j.neunet.2026.108797_bib0050","series-title":"International conference on machine learning","first-page":"11932","article-title":"Elementary superexpressive activations","author":"Yarotsky","year":"2021"},{"issue":"35","key":"10.1016\/j.neunet.2026.108797_bib0051","first-page":"1","article-title":"Deep network approximation: Beyond ReLU to diverse activation functions","volume":"25","author":"Zhang","year":"2024","journal-title":"Journal of Machine Learning Research"},{"issue":"276","key":"10.1016\/j.neunet.2026.108797_bib0052","first-page":"1","article-title":"Deep network approximation: Achieving arbitrary accuracy with fixed number of neurons","volume":"23","author":"Zhang","year":"2022","journal-title":"Journal of Machine Learning Research"},{"key":"10.1016\/j.neunet.2026.108797_bib0053","first-page":"5669","article-title":"Neural network architecture beyond width and depth","volume":"35","author":"Zhang","year":"2022","journal-title":"Advances in Neural Information Processing Systems"},{"issue":"1","key":"10.1016\/j.neunet.2026.108797_bib0054","article-title":"Physics-informed neural networks with hybrid Kolmogorov-Arnold network and augmented Lagrangian function for solving partial differential equations","volume":"15","author":"Zhang","year":"2025","journal-title":"Scientific Reports"}],"container-title":["Neural Networks"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/api.elsevier.com\/content\/article\/PII:S0893608026002595?httpAccept=text\/xml","content-type":"text\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/api.elsevier.com\/content\/article\/PII:S0893608026002595?httpAccept=text\/plain","content-type":"text\/plain","content-version":"vor","intended-application":"text-mining"}],"deposited":{"date-parts":[[2026,5,25]],"date-time":"2026-05-25T06:34:13Z","timestamp":1779690853000},"score":1,"resource":{"primary":{"URL":"https:\/\/linkinghub.elsevier.com\/retrieve\/pii\/S0893608026002595"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8]]},"references-count":54,"alternative-id":["S0893608026002595"],"URL":"https:\/\/doi.org\/10.1016\/j.neunet.2026.108797","relation":{},"ISSN":["0893-6080"],"issn-type":[{"value":"0893-6080","type":"print"}],"subject":[],"published":{"date-parts":[[2026,8]]},"assertion":[{"value":"Elsevier","name":"publisher","label":"This article is maintained by"},{"value":"Approximation rates in Besov norms and sample-complexity of Kolmogorov-Arnold networks with residual connections","name":"articletitle","label":"Article Title"},{"value":"Neural Networks","name":"journaltitle","label":"Journal Title"},{"value":"https:\/\/doi.org\/10.1016\/j.neunet.2026.108797","name":"articlelink","label":"CrossRef DOI link to publisher maintained version"},{"value":"article","name":"content_type","label":"Content Type"},{"value":"\u00a9 2026 Published by Elsevier Ltd.","name":"copyright","label":"Copyright"}],"article-number":"108797"}}