{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,9]],"date-time":"2026-10-09T04:17:18Z","timestamp":1791519438406,"version":"4.3.4"},"reference-count":212,"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":[[2026,3,11]],"date-time":"2026-03-11T00:00:00Z","timestamp":1773187200000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/100000006","name":"Office of Naval Research","doi-asserted-by":"publisher","award":["N00014-23-1-2394"],"award-info":[{"award-number":["N00014-23-1-2394"]}],"id":[{"id":"10.13039\/100000006","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000006","name":"Office of Naval Research","doi-asserted-by":"publisher","award":["N00014-23-1-2790"],"award-info":[{"award-number":["N00014-23-1-2790"]}],"id":[{"id":"10.13039\/100000006","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.2026.108841","type":"journal-article","created":{"date-parts":[[2026,3,13]],"date-time":"2026-03-13T16:32:16Z","timestamp":1773419536000},"page":"108841","update-policy":"https:\/\/doi.org\/10.1016\/elsevier_cm_policy","source":"Crossref","is-referenced-by-count":5,"special_numbering":"C","title":["An approximation theory perspective on machine learning"],"prefix":"10.1016","volume":"200","author":[{"given":"Hrushikesh N.","family":"Mhaskar","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Efstratios","family":"Tsoukanis","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8831-1000","authenticated-orcid":false,"given":"Ameya D.","family":"Jagtap","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"78","reference":[{"key":"10.1016\/j.neunet.2026.108841_bib0001","doi-asserted-by":"crossref","DOI":"10.1016\/j.neucom.2025.131440","article-title":"Challenges and advancements in modeling shock fronts with physics-informed neural networks: A review and benchmarking study","volume":"657","author":"Abbasi","year":"2025","journal-title":"Neurocomputing"},{"key":"10.1016\/j.neunet.2026.108841_bib0002","doi-asserted-by":"crossref","DOI":"10.1016\/j.cma.2025.117784","article-title":"History-matching of imbibition flow in fractured porous media using physics-informed neural networks (PINNs)","volume":"437","author":"Abbasi","year":"2025","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"key":"10.1016\/j.neunet.2026.108841_bib0003","series-title":"Learning from data","volume":"vol. 4","author":"Abu-Mostafa","year":"2012"},{"key":"10.1016\/j.neunet.2026.108841_bib0004","unstructured":"Adcock, B., Brugiapaglia, S., Dexter, N., & Moraga, S. (2020). Deep neural networks are effective at learning high-dimensional hilbert-valued functions from limited data. arXiv preprint arXiv: 2012.06081."},{"key":"10.1016\/j.neunet.2026.108841_bib0005","series-title":"AAAI spring symposium: MLPs","article-title":"Learning high-dimensional hilbert-valued functions with deep neural networks from limited data","author":"Adcock","year":"2021"},{"key":"10.1016\/j.neunet.2026.108841_bib0006","series-title":"Neural network learning: Theoretical foundations","author":"Anthony","year":"2009"},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0007","doi-asserted-by":"crossref","first-page":"337","DOI":"10.1090\/S0002-9947-1950-0051437-7","article-title":"Theory of reproducing kernels","volume":"68","author":"Aronszajn","year":"1950","journal-title":"Transactions of the American Mathematical Society"},{"issue":"20","key":"10.1016\/j.neunet.2026.108841_bib0008","doi-asserted-by":"crossref","first-page":"11767","DOI":"10.1007\/s00500-024-09918-2","article-title":"On the accuracy of interpolation based on single-layer artificial neural networks with a focus on defeating the runge phenomenon","volume":"28","author":"Auricchio","year":"2024","journal-title":"Soft Computing"},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0009","doi-asserted-by":"crossref","first-page":"930","DOI":"10.1109\/18.256500","article-title":"Universal approximation bounds for superpositions of a sigmoidal function","volume":"39","author":"Barron","year":"1993","journal-title":"IEEE Transactions on Information Theory"},{"key":"10.1016\/j.neunet.2026.108841_bib0010","doi-asserted-by":"crossref","first-page":"203","DOI":"10.1017\/S0962492921000039","article-title":"Fit without fear: Remarkable mathematical phenomena of deep learning through the prism of interpolation","volume":"30","author":"Belkin","year":"2021","journal-title":"Acta Numerica"},{"issue":"32","key":"10.1016\/j.neunet.2026.108841_bib0011","doi-asserted-by":"crossref","first-page":"15849","DOI":"10.1073\/pnas.1903070116","article-title":"Reconciling modern machine-learning practice and the classical bias\u2013variance trade-off","volume":"116","author":"Belkin","year":"2019","journal-title":"Proceedings of the National Academy of Sciences"},{"issue":"4","key":"10.1016\/j.neunet.2026.108841_bib0012","doi-asserted-by":"crossref","first-page":"1167","DOI":"10.1137\/20M1336072","article-title":"Two models of double descent for weak features","volume":"2","author":"Belkin","year":"2020","journal-title":"SIAM Journal on Mathematics of Data Science"},{"key":"10.1016\/j.neunet.2026.108841_bib0013","article-title":"Convergence of Laplacian eigenmaps","volume":"129","author":"Belkin","year":"2007","journal-title":"Advances in Neural Information Processing 19"},{"issue":"8","key":"10.1016\/j.neunet.2026.108841_bib0014","doi-asserted-by":"crossref","first-page":"1289","DOI":"10.1016\/j.jcss.2007.08.006","article-title":"Towards a theoretical foundation for Laplacian-based manifold methods","volume":"74","author":"Belkin","year":"2008","journal-title":"Journal of Computer and System Sciences"},{"key":"10.1016\/j.neunet.2026.108841_bib0015","series-title":"Pattern recognition and machine learning","volume":"vol. 4","author":"Bishop","year":"2006"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0016","doi-asserted-by":"crossref","first-page":"8","DOI":"10.1137\/18M118709X","article-title":"Optimal approximation with sparsely connected deep neural networks","volume":"1","author":"Bolcskei","year":"2019","journal-title":"SIAM Journal on Mathematics of Data Science"},{"key":"10.1016\/j.neunet.2026.108841_bib0017","unstructured":"Burkholz, R., Laha, N., Mukherjee, R., & Gotovos, A. (2021). On the existence of universal lottery tickets. arXiv preprint arXiv: 2111.11146."},{"key":"10.1016\/j.neunet.2026.108841_bib0018","unstructured":"Cao, Y., Fang, Z., Wu, Y., Zhou, D.-X., & Gu, Q. (2019). Towards understanding the spectral bias of deep learning. arXiv preprint arXiv: 1912.01198."},{"key":"10.1016\/j.neunet.2026.108841_bib0019","doi-asserted-by":"crossref","first-page":"149","DOI":"10.1016\/j.jcp.2018.03.014","article-title":"Minimum Sobolev norm interpolation of scattered derivative data","volume":"365","author":"Chandrasekaran","year":"2018","journal-title":"Journal of Computational Physics"},{"key":"10.1016\/j.neunet.2026.108841_bib0020","doi-asserted-by":"crossref","first-page":"96","DOI":"10.1016\/j.jcp.2013.03.041","article-title":"Minimum Sobolev norm interpolation with trigonometric polynomials on the torus","volume":"249","author":"Chandrasekaran","year":"2013","journal-title":"Journal of Computational Physics"},{"key":"10.1016\/j.neunet.2026.108841_bib0021","unstructured":"Chen, K., Wang, C., & Yang, H. (2023). Deep operator learning lessens the curse of dimensionality for PDEs. arXiv preprint arXiv: 2301.12227."},{"issue":"6","key":"10.1016\/j.neunet.2026.108841_bib0022","doi-asserted-by":"crossref","first-page":"910","DOI":"10.1109\/72.286886","article-title":"Approximations of continuous functionals by neural networks with application to dynamic systems","volume":"4","author":"Chen","year":"1993","journal-title":"IEEE Transactions on Neural Networks"},{"issue":"4","key":"10.1016\/j.neunet.2026.108841_bib0023","doi-asserted-by":"crossref","first-page":"911","DOI":"10.1109\/72.392253","article-title":"Universal approximation to nonlinear operators by neural networks with arbitrary activation functions and its application to dynamical systems","volume":"6","author":"Chen","year":"1995","journal-title":"IEEE Transactions on Neural Networks"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0024","doi-asserted-by":"crossref","DOI":"10.1016\/j.acha.2006.05.005","article-title":"Special issue: Diffusion maps and wavelets","volume":"21","author":"Chui","year":"2006","journal-title":"Applied and Computational Harmonic Analysis"},{"issue":"2","key":"10.1016\/j.neunet.2026.108841_bib0025","doi-asserted-by":"crossref","first-page":"131","DOI":"10.1016\/0021-9045(92)90081-X","article-title":"Approximation by ridge functions and neural networks with one hidden layer","volume":"70","author":"Chui","year":"1992","journal-title":"Journal of Approximation Theory"},{"key":"10.1016\/j.neunet.2026.108841_bib0026","first-page":"77","article-title":"Realization of neural networks with one hidden layer","author":"Chui","year":"1993","journal-title":"Multivariate Approximation: From CAGD to Wavelets, World Scientific, Singapore"},{"issue":"208","key":"10.1016\/j.neunet.2026.108841_bib0027","doi-asserted-by":"crossref","first-page":"607","DOI":"10.1090\/S0025-5718-1994-1240656-2","article-title":"Neural networks for localized approximation","volume":"63","author":"Chui","year":"1994","journal-title":"Mathematics of Computation"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0028","doi-asserted-by":"crossref","first-page":"233","DOI":"10.1007\/BF02124745","article-title":"Limitations of the approximation capabilities of neural networks with one hidden layer","volume":"5","author":"Chui","year":"1996","journal-title":"Advances in Computational Mathematics"},{"key":"10.1016\/j.neunet.2026.108841_bib0029","doi-asserted-by":"crossref","first-page":"12","DOI":"10.3389\/fams.2018.00012","article-title":"Deep nets for local manifold learning","volume":"4","author":"Chui","year":"2018","journal-title":"Frontiers in Applied Mathematics and Statistics"},{"key":"10.1016\/j.neunet.2026.108841_bib0030","series-title":"Spectral graph theory","volume":"vol. 92","author":"Chung","year":"1997"},{"key":"10.1016\/j.neunet.2026.108841_bib0031","doi-asserted-by":"crossref","first-page":"404","DOI":"10.1016\/j.neunet.2021.06.004","article-title":"A deep network construction that adapts to intrinsic dimensionality beyond the domain","volume":"141","author":"Cloninger","year":"2021","journal-title":"Neural Networks"},{"key":"10.1016\/j.neunet.2026.108841_bib0032","doi-asserted-by":"crossref","first-page":"44","DOI":"10.1016\/j.acha.2021.02.002","article-title":"Cautious active clustering","volume":"54","author":"Cloninger","year":"2021","journal-title":"Applied and Computational Harmonic Analysis"},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0033","doi-asserted-by":"crossref","first-page":"607","DOI":"10.1007\/s10208-021-09494-z","article-title":"Optimal stable nonlinear approximation","volume":"22","author":"Cohen","year":"2022","journal-title":"Foundations of Computational Mathematics"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0034","doi-asserted-by":"crossref","first-page":"5","DOI":"10.1016\/j.acha.2006.04.006","article-title":"Diffusion maps","volume":"21","author":"Coifman","year":"2006","journal-title":"Applied and Computational Harmonic Analysis"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0035","first-page":"26","article-title":"Condiciones para que una funcion infinitamente derivable sea un polinomio","volume":"14","author":"Corominas","year":"1954","journal-title":"Revista Matem\u00e1tica Hispanoamericana"},{"key":"10.1016\/j.neunet.2026.108841_bib0036","series-title":"Learning theory: An approximation theory viewpoint","volume":"vol. 24","author":"Cucker","year":"2007"},{"issue":"4","key":"10.1016\/j.neunet.2026.108841_bib0037","doi-asserted-by":"crossref","first-page":"303","DOI":"10.1007\/BF02551274","article-title":"Approximation by superposition of sigmoidal functions","volume":"2","author":"Cybenko","year":"1989","journal-title":"Mathematics of Control, Signals and Systems"},{"key":"10.1016\/j.neunet.2026.108841_bib0038","series-title":"Approximation theory and harmonic analysis on spheres and balls","volume":"vol. 23","author":"Dai","year":"2013"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0039","doi-asserted-by":"crossref","first-page":"83","DOI":"10.1093\/imanum\/drac085","article-title":"Error estimates for physics-informed neural networks approximating the navier\u2013stokes equations","volume":"44","author":"De Ryck","year":"2024","journal-title":"IMA Journal of Numerical Analysis"},{"issue":"6","key":"10.1016\/j.neunet.2026.108841_bib0040","doi-asserted-by":"crossref","first-page":"79","DOI":"10.1007\/s10444-022-09985-9","article-title":"Error analysis for physics-informed neural networks (PINNs) approximating Kolmogorov PDEs","volume":"48","author":"De Ryck","year":"2022","journal-title":"Advances in Computational Mathematics"},{"key":"10.1016\/j.neunet.2026.108841_bib0041","doi-asserted-by":"crossref","first-page":"633","DOI":"10.1017\/S0962492923000089","article-title":"Numerical analysis of physics-informed neural networks and related models in physics-informed machine learning","volume":"33","author":"De Ryck","year":"2024","journal-title":"Acta Numerica"},{"issue":"4","key":"10.1016\/j.neunet.2026.108841_bib0042","doi-asserted-by":"crossref","first-page":"469","DOI":"10.1007\/BF01171759","article-title":"Optimal nonlinear approximation","volume":"63","author":"DeVore","year":"1989","journal-title":"Manuscripta Mathematica"},{"key":"10.1016\/j.neunet.2026.108841_bib0043","series-title":"Constructive approximation","volume":"vol. 303","author":"DeVore","year":"1993"},{"issue":"4","key":"10.1016\/j.neunet.2026.108841_bib0044","doi-asserted-by":"crossref","first-page":"489","DOI":"10.1006\/jcom.1997.0458","article-title":"Nonlinear approximation in finite-dimensional spaces","volume":"13","author":"DeVore","year":"1997","journal-title":"Journal of Complexity"},{"key":"10.1016\/j.neunet.2026.108841_bib0045","article-title":"Neural operator prediction of linear instability waves in high-speed boundary layers","volume":"474","author":"Di Leoni","year":"2023","journal-title":"Journal of Computational Physics"},{"issue":"11","key":"10.1016\/j.neunet.2026.108841_bib0046","doi-asserted-by":"crossref","first-page":"1251","DOI":"10.1089\/cmb.2012.0187","article-title":"Locally learning biomedical data using diffusion frames","volume":"19","author":"Ehler","year":"2012","journal-title":"Journal of Computational Biology"},{"key":"10.1016\/j.neunet.2026.108841_bib0047","series-title":"Doklady mathematics","first-page":"S226","article-title":"Spectral neural operators","volume":"vol. 108","author":"Fanaskov","year":"2023"},{"key":"10.1016\/j.neunet.2026.108841_bib0048","doi-asserted-by":"crossref","unstructured":"Fang, R., & Xu, Y. (2024). Addressing spectral bias of deep neural networks by multi-grade deep learning. arXiv preprint arXiv: 2410.16105.","DOI":"10.52202\/079017-3625"},{"key":"10.1016\/j.neunet.2026.108841_bib0049","first-page":"1","article-title":"Fitting a manifold of large reach to noisy data","author":"Fefferman","year":"2023","journal-title":"Journal of Topology and Analysis"},{"issue":"4","key":"10.1016\/j.neunet.2026.108841_bib0050","doi-asserted-by":"crossref","first-page":"983","DOI":"10.1090\/jams\/852","article-title":"Testing the manifold hypothesis","volume":"29","author":"Fefferman","year":"2016","journal-title":"Journal of the American Mathematical Society"},{"issue":"5","key":"10.1016\/j.neunet.2026.108841_bib0051","doi-asserted-by":"crossref","first-page":"629","DOI":"10.1007\/s00041-010-9119-4","article-title":"A quadrature formula for diffusion polynomials corresponding to a generalized heat kernel","volume":"16","author":"Filbir","year":"2010","journal-title":"Journal of Fourier Analysis and Applications"},{"issue":"6","key":"10.1016\/j.neunet.2026.108841_bib0052","doi-asserted-by":"crossref","first-page":"568","DOI":"10.1016\/j.jco.2011.03.002","article-title":"Marcinkiewicz\u2013Zygmund measures on manifolds","volume":"27","author":"Filbir","year":"2011","journal-title":"Journal of Complexity"},{"issue":"5","key":"10.1016\/j.neunet.2026.108841_bib0053","doi-asserted-by":"crossref","first-page":"650","DOI":"10.1002\/mana.200710633","article-title":"Polynomial approximation on the sphere using scattered data","volume":"281","author":"Filbir","year":"2008","journal-title":"Mathematische Nachrichten"},{"key":"10.1016\/j.neunet.2026.108841_bib0054","unstructured":"Filom, K., Kording, K. P., & Farhoodi, R. (2020). Pde constraints on smooth hierarchical functions computed by neural networks. arXiv preprint arXiv: 2005.08859."},{"key":"10.1016\/j.neunet.2026.108841_bib0055","doi-asserted-by":"crossref","first-page":"7368","DOI":"10.52202\/068431-0535","article-title":"Spectral bias in practice: The role of function frequency in generalization","volume":"35","author":"Fridovich-Keil","year":"2022","journal-title":"Advances in Neural Information Processing Systems"},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0056","doi-asserted-by":"crossref","first-page":"183","DOI":"10.1016\/0893-6080(89)90003-8","article-title":"On the approximate realization of continuous mappings by neural networks","volume":"2","author":"Funahashi","year":"1989","journal-title":"Neural Networks"},{"issue":"2","key":"10.1016\/j.neunet.2026.108841_bib0057","doi-asserted-by":"crossref","first-page":"334","DOI":"10.1007\/s12220-010-9150-3","article-title":"Band-limited localized Parseval frames and Besov spaces on compact homogeneous manifolds","volume":"21","author":"Geller","year":"2011","journal-title":"Journal of Geometric Analysis"},{"key":"10.1016\/j.neunet.2026.108841_bib0058","unstructured":"Geshkovski, B., Letrouit, C., Polyanskiy, Y., & Rigollet, P. (2023). A mathematical perspective on transformers. arXiv preprint arXiv: 2312.10794."},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0059","doi-asserted-by":"crossref","first-page":"169","DOI":"10.1007\/BF00195855","article-title":"Networks and the best approximation property","volume":"63","author":"Girosi","year":"1990","journal-title":"Biological Cybernetics"},{"key":"10.1016\/j.neunet.2026.108841_bib0060","first-page":"117","article-title":"Optimal approximation and error bounds","author":"Golomb","year":"1959","journal-title":"On Numerical Approximation"},{"key":"10.1016\/j.neunet.2026.108841_bib0061","series-title":"Deep learning","volume":"vol. 1","author":"Goodfellow","year":"2016"},{"key":"10.1016\/j.neunet.2026.108841_bib0062","doi-asserted-by":"crossref","DOI":"10.1016\/j.cma.2023.116674","article-title":"Learning stiff chemical kinetics using extended deep neural operators","volume":"419","author":"Goswami","year":"2024","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0063","doi-asserted-by":"crossref","first-page":"124","DOI":"10.1016\/j.acha.2008.12.003","article-title":"On the computation of nonnegative quadrature weights on the sphere","volume":"27","author":"Gr\u00e4f","year":"2009","journal-title":"Applied and Computational Harmonic Analysis"},{"issue":"7","key":"10.1016\/j.neunet.2026.108841_bib0064","doi-asserted-by":"crossref","first-page":"1289","DOI":"10.1162\/NECO_a_00849","article-title":"A single hidden layer feedforward network with only one neuron in the hidden layer can approximate any univariate function","volume":"28","author":"Guliyev","year":"2016","journal-title":"Neural Computation"},{"key":"10.1016\/j.neunet.2026.108841_bib0065","series-title":"Proceedings of the 22nd international conference on machine learning","first-page":"289","article-title":"Intrinsic dimensionality estimation of submanifolds in rd","author":"Hein","year":"2005"},{"key":"10.1016\/j.neunet.2026.108841_bib0066","series-title":"2022 International joint conference on neural networks (IJCNN)","first-page":"1","article-title":"Certified machine learning: A posteriori error estimation for physics-informed neural networks","author":"Hillebrecht","year":"2022"},{"key":"10.1016\/j.neunet.2026.108841_bib0067","series-title":"Matrix analysis","author":"Horn","year":"2012"},{"issue":"5","key":"10.1016\/j.neunet.2026.108841_bib0068","doi-asserted-by":"crossref","first-page":"359","DOI":"10.1016\/0893-6080(89)90020-8","article-title":"Multilayer feedforward networks are universal approximators","volume":"2","author":"Hornik","year":"1989","journal-title":"Neural Networks"},{"issue":"5","key":"10.1016\/j.neunet.2026.108841_bib0069","doi-asserted-by":"crossref","first-page":"A3158","DOI":"10.1137\/21M1447039","article-title":"When do extended physics-informed neural networks (XPINNs) improve generalization?","volume":"44","author":"Hu","year":"2022","journal-title":"SIAM Journal on Scientific Computing"},{"key":"10.1016\/j.neunet.2026.108841_bib0070","doi-asserted-by":"crossref","first-page":"32","DOI":"10.1016\/j.neunet.2014.10.001","article-title":"Trends in extreme learning machines: A review","volume":"61","author":"Huang","year":"2015","journal-title":"Neural Networks"},{"key":"10.1016\/j.neunet.2026.108841_bib0071","series-title":"Neural networks, 1988., IEEE international conference on","first-page":"641","article-title":"Capabilities of three-layered perceptrons","author":"Irie","year":"1988"},{"key":"10.1016\/j.neunet.2026.108841_bib0072","series-title":"Wavelets and sparsity XVI","first-page":"336","article-title":"Applied and computational harmonic analysis on graphs and networks","volume":"vol. 9597","author":"Irion","year":"2015"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0073","doi-asserted-by":"crossref","first-page":"189","DOI":"10.1007\/BF02124743","article-title":"Nonlinearity creates linear independence","volume":"5","author":"Ito","year":"1996","journal-title":"Advances in Computational Mathematics"},{"issue":"5","key":"10.1016\/j.neunet.2026.108841_bib0074","doi-asserted-by":"crossref","first-page":"2002","DOI":"10.4208\/cicp.OA-2020-0164","article-title":"Extended physics-informed neural networks (XPINNs): A generalized space-time domain decomposition based deep learning framework for nonlinear partial differential equations","volume":"28","author":"Jagtap","year":"2020","journal-title":"Communications in Computational Physics"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0075","doi-asserted-by":"crossref","first-page":"21","DOI":"10.1615\/JMachLearnModelComput.2023047367","article-title":"How important are activation functions in regression and classification? A survey, performance comparison, and future directions","volume":"4","author":"Jagtap","year":"2023","journal-title":"Journal of Machine Learning for Modeling and Computing"},{"issue":"2239","key":"10.1016\/j.neunet.2026.108841_bib0076","article-title":"Locally adaptive activation functions with slope recovery for deep and physics-informed neural networks","volume":"476","author":"Jagtap","year":"2020","journal-title":"Proceedings of the Royal Society A"},{"key":"10.1016\/j.neunet.2026.108841_bib0077","doi-asserted-by":"crossref","DOI":"10.1016\/j.jcp.2019.109136","article-title":"Adaptive activation functions accelerate convergence in deep and physics-informed neural networks","volume":"404","author":"Jagtap","year":"2020","journal-title":"Journal of Computational Physics"},{"key":"10.1016\/j.neunet.2026.108841_bib0078","doi-asserted-by":"crossref","DOI":"10.1016\/j.cma.2020.113028","article-title":"Conservative physics-informed neural networks on discrete domains for conservation laws: Applications to forward and inverse problems","volume":"365","author":"Jagtap","year":"2020","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"key":"10.1016\/j.neunet.2026.108841_bib0079","doi-asserted-by":"crossref","DOI":"10.1016\/j.jcp.2022.111402","article-title":"Physics-informed neural networks for inverse problems in supersonic flows","volume":"466","author":"Jagtap","year":"2022","journal-title":"Journal of Computational Physics"},{"key":"10.1016\/j.neunet.2026.108841_bib0080","doi-asserted-by":"crossref","DOI":"10.1016\/j.oceaneng.2022.110775","article-title":"Deep learning of inverse water waves problems using multi-fidelity data: Application to Serre\u2013Green\u2013Naghdi equations","volume":"248","author":"Jagtap","year":"2022","journal-title":"Ocean Engineering"},{"key":"10.1016\/j.neunet.2026.108841_bib0081","doi-asserted-by":"crossref","first-page":"165","DOI":"10.1016\/j.neucom.2021.10.036","article-title":"Deep kronecker neural networks: A general framework for neural networks with adaptive activation functions","volume":"468","author":"Jagtap","year":"2022","journal-title":"Neurocomputing"},{"key":"10.1016\/j.neunet.2026.108841_bib0082","doi-asserted-by":"crossref","first-page":"85","DOI":"10.1016\/j.neunet.2020.06.024","article-title":"Quantifying the generalization error in deep learning in terms of data distribution and neural network smoothness","volume":"130","author":"Jin","year":"2020","journal-title":"Neural Networks"},{"key":"10.1016\/j.neunet.2026.108841_bib0083","doi-asserted-by":"crossref","first-page":"131","DOI":"10.5186\/aasfm.2010.3508","article-title":"Universal local parametrizations via heat kernels and eigenfunctions of the Laplacian","volume":"35","author":"Jones","year":"2010","journal-title":"Annales Academi\u00e6 Scientiarum Fennic\u00e6 Mathematica"},{"issue":"4","key":"10.1016\/j.neunet.2026.108841_bib0084","doi-asserted-by":"crossref","first-page":"19","DOI":"10.1145\/1555386.1555388","article-title":"Using NFFT 3\u2014a software library for various nonequispaced fast fourier transforms","volume":"36","author":"Keiner","year":"2009","journal-title":"ACM Transactions on Mathematical Software (TOMS)"},{"key":"10.1016\/j.neunet.2026.108841_bib0085","doi-asserted-by":"crossref","DOI":"10.1016\/j.cma.2020.113547","article-title":"hp-VPINNs: Variational physics-informed neural networks with domain decomposition","volume":"374","author":"Kharazmi","year":"2021","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"key":"10.1016\/j.neunet.2026.108841_bib0086","doi-asserted-by":"crossref","first-page":"49","DOI":"10.1016\/j.neunet.2023.04.014","article-title":"Theoretical bounds of generalization error for generalized extreme learning machine and random vector functional link network","volume":"164","author":"Kim","year":"2023","journal-title":"Neural Networks"},{"issue":"12","key":"10.1016\/j.neunet.2026.108841_bib0087","doi-asserted-by":"crossref","first-page":"7649","DOI":"10.1109\/TIT.2018.2874447","article-title":"Approximation by combinations of reLU and squared reLU ridge functions with \u21131 and \u21130 controls","volume":"64","author":"Klusowski","year":"2018","journal-title":"IEEE Transactions on Information Theory"},{"issue":"6","key":"10.1016\/j.neunet.2026.108841_bib0088","doi-asserted-by":"crossref","first-page":"4032","DOI":"10.1109\/TIT.2022.3146620","article-title":"Estimation of a function of low local dimensionality by deep neural networks","volume":"68","author":"Kohler","year":"2022","journal-title":"IEEE Transactions on Information Theory"},{"issue":"4","key":"10.1016\/j.neunet.2026.108841_bib0089","doi-asserted-by":"crossref","first-page":"2231","DOI":"10.1214\/20-AOS2034","article-title":"On the rate of convergence of fully connected deep neural network regression estimates","volume":"49","author":"Kohler","year":"2021","journal-title":"The Annals of Statistics"},{"key":"10.1016\/j.neunet.2026.108841_bib0090","doi-asserted-by":"crossref","first-page":"160","DOI":"10.1016\/j.jspi.2022.05.008","article-title":"Estimation of a regression function on a manifold by fully connected deep neural networks","volume":"222","author":"Kohler","year":"2023","journal-title":"Journal of Statistical Planning and Inference"},{"key":"10.1016\/j.neunet.2026.108841_bib0091","article-title":"On universal approximation and error bounds for fourier neural operators","volume":"22","author":"Kovachki","year":"2021","journal-title":"Journal of Machine Learning Research"},{"key":"10.1016\/j.neunet.2026.108841_bib0092","first-page":"26548","article-title":"Characterizing possible failure modes in physics-informed neural networks","volume":"34","author":"Krishnapriyan","year":"2021","journal-title":"Advances in Neural Information Processing Systems"},{"issue":"6","key":"10.1016\/j.neunet.2026.108841_bib0093","doi-asserted-by":"crossref","first-page":"2659","DOI":"10.1109\/18.945285","article-title":"Bounds on rates of variable basis and neural network approximation","volume":"47","author":"K\u016frkov\u00e1","year":"2001","journal-title":"IEEE Transactions on Information Theory"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0094","doi-asserted-by":"crossref","first-page":"264","DOI":"10.1109\/18.971754","article-title":"Comparison of worst case errors in linear and neural network approximation","volume":"48","author":"K\u016frkov\u00e1","year":"2002","journal-title":"IEEE Transactions on Information Theory"},{"issue":"9","key":"10.1016\/j.neunet.2026.108841_bib0095","doi-asserted-by":"crossref","first-page":"1393","DOI":"10.1109\/TPAMI.2006.184","article-title":"Diffusion maps and coarse-graining: A unified framework for dimensionality reduction, graph partitioning, and data set parameterization","volume":"28","author":"Lafon","year":"2006","journal-title":"IEEE transactions on pattern analysis and machine intelligence"},{"key":"10.1016\/j.neunet.2026.108841_bib0096","series-title":"Diffusion maps and geometric harmonics","author":"Lafon","year":"2004"},{"key":"10.1016\/j.neunet.2026.108841_bib0097","unstructured":"Lai, M.-J., & Shen, Z. (2021). The Kolmogorov superposition theorem can break the curse of dimensionality when approximating high dimensional functions. arXiv preprint arXiv: 2112.09963."},{"key":"10.1016\/j.neunet.2026.108841_bib0098","unstructured":"Lai, M.-J., & Shen, Z. (2024). The optimal rate for linear KB-splines and LKB-splines approximation of high dimensional continuous functions and its application. arXiv preprint arXiv: 2401.03956."},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0099","doi-asserted-by":"crossref","DOI":"10.1093\/imatrm\/tnac001","article-title":"Error estimates for deeponets: A deep learning framework in infinite dimensions","volume":"6","author":"Lanthaler","year":"2022","journal-title":"Transactions of Mathematics and Its Applications"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0100","doi-asserted-by":"crossref","first-page":"440","DOI":"10.1137\/060678555","article-title":"Localized linear polynomial operators and quadrature formulas on the sphere","volume":"47","author":"Le Gia","year":"2009","journal-title":"SIAM Journal on Numerical Analysis"},{"issue":"6","key":"10.1016\/j.neunet.2026.108841_bib0101","doi-asserted-by":"crossref","first-page":"861","DOI":"10.1016\/S0893-6080(05)80131-5","article-title":"Multilayer feedforward networks with a nonpolynomial activation function can approximate any function","volume":"6","author":"Leshno","year":"1993","journal-title":"Neural Networks"},{"key":"10.1016\/j.neunet.2026.108841_bib0102","series-title":"International conference on machine learning","first-page":"19565","article-title":"Transformers as algorithms: Generalization and stability in-context learning","author":"Li","year":"2023"},{"key":"10.1016\/j.neunet.2026.108841_bib0103","unstructured":"Li, Z., Kovachki, N., Azizzadenesheli, K., Liu, B., Bhattacharya, K., Stuart, A., & Anandkumar, A. (2020). Fourier neural operator for parametric partial differential equations. arXiv preprint arXiv: 2010.08895."},{"issue":"24","key":"10.1016\/j.neunet.2026.108841_bib0104","first-page":"1","article-title":"Deep nonparametric estimation of operators between infinite dimensional spaces","volume":"25","author":"Liu","year":"2024","journal-title":"Journal of Machine Learning Research"},{"key":"10.1016\/j.neunet.2026.108841_bib0105","unstructured":"Liu, Z., Wang, Y., Vaidya, S., Ruehle, F., Halverson, J., Solja\u010di\u0107, M., Hou, T. Y., & Tegmark, M. (2024b). Kan: Kolmogorov-Arnold networks. arXiv preprint arXiv: 2404.19756."},{"key":"10.1016\/j.neunet.2026.108841_bib0106","series-title":"Approximation of functions","volume":"vol. 322","author":"Lorentz","year":"2005"},{"key":"10.1016\/j.neunet.2026.108841_bib0107","series-title":"Constructive approximation: advanced problems","volume":"vol. 304","author":"Lorentz","year":"1996"},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0108","doi-asserted-by":"crossref","first-page":"218","DOI":"10.1038\/s42256-021-00302-5","article-title":"Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators","volume":"3","author":"Lu","year":"2021","journal-title":"Nature Machine Intelligence"},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0109","doi-asserted-by":"crossref","first-page":"46","DOI":"10.1007\/s40687-022-00346-y","article-title":"Uniform approximation rates and metric entropy of shallow neural networks","volume":"9","author":"Ma","year":"2022","journal-title":"Research in the Mathematical Sciences"},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0110","doi-asserted-by":"crossref","first-page":"329","DOI":"10.1016\/j.acha.2007.07.001","article-title":"Diffusion polynomial frames on metric measure spaces","volume":"24","author":"Maggioni","year":"2008","journal-title":"Applied and Computational Harmonic Analysis"},{"issue":"2","key":"10.1016\/j.neunet.2026.108841_bib0111","doi-asserted-by":"crossref","first-page":"215","DOI":"10.1006\/jath.1997.3217","article-title":"Uniform approximation by neural networks","volume":"95","author":"Makovoz","year":"1998","journal-title":"Journal of Approximation Theory"},{"key":"10.1016\/j.neunet.2026.108841_bib0112","doi-asserted-by":"crossref","first-page":"778","DOI":"10.1016\/j.neunet.2021.09.027","article-title":"Theory of deep convolutional neural networks III: Approximating radial functions","volume":"144","author":"Mao","year":"2021","journal-title":"Neural Networks"},{"issue":"01","key":"10.1016\/j.neunet.2026.108841_bib0113","doi-asserted-by":"crossref","first-page":"93","DOI":"10.1142\/S0219530522400085","article-title":"Approximating functions with multi-features by deep convolutional neural networks","volume":"21","author":"Mao","year":"2023","journal-title":"Analysis and Applications"},{"key":"10.1016\/j.neunet.2026.108841_bib0114","doi-asserted-by":"crossref","DOI":"10.1016\/j.cma.2019.112789","article-title":"Physics-informed neural networks for high-speed flows","volume":"360","author":"Mao","year":"2020","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"key":"10.1016\/j.neunet.2026.108841_bib0115","doi-asserted-by":"crossref","first-page":"353","DOI":"10.1016\/j.neunet.2022.04.024","article-title":"A manifold learning approach for gesture identification from micro-doppler radar measurements","volume":"152","author":"Mason","year":"2022","journal-title":"Neural Networks"},{"key":"10.1016\/j.neunet.2026.108841_bib0116","doi-asserted-by":"crossref","DOI":"10.1016\/j.cma.2025.118403","article-title":"Anant-Net: Breaking the curse of dimensionality with scalable and interpretable neural surrogate for high-dimensional PDEs","volume":"447","author":"Menon","year":"2025","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"key":"10.1016\/j.neunet.2026.108841_bib0117","doi-asserted-by":"crossref","DOI":"10.1016\/j.neunet.2026.108567","article-title":"On scientific foundation models: Rigorous definitions, key applications, and a comprehensive survey","volume":"198","author":"Menon","year":"2026","journal-title":"Neural Networks"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0118","doi-asserted-by":"crossref","first-page":"61","DOI":"10.1007\/BF02070821","article-title":"Approximation properties of a multilayered feedforward artificial neural network","volume":"1","author":"Mhaskar","year":"1993","journal-title":"Advances in Computational Mathematics"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0119","doi-asserted-by":"crossref","first-page":"164","DOI":"10.1162\/neco.1996.8.1.164","article-title":"Neural networks for optimal approximation of smooth and analytic functions","volume":"8","author":"Mhaskar","year":"1996","journal-title":"Neural Computation"},{"key":"10.1016\/j.neunet.2026.108841_bib0120","doi-asserted-by":"crossref","first-page":"275","DOI":"10.1007\/978-1-4615-6099-9_47","article-title":"On smooth activation functions","author":"Mhaskar","year":"1997","journal-title":"Mathematics of Neural Networks: Models, Algorithms and Applications"},{"key":"10.1016\/j.neunet.2026.108841_bib0121","series-title":"Wavelet analysis and applications, proceedings of the international workshop in Delhi","first-page":"247","article-title":"Approximation theory and neural networks","author":"Mhaskar","year":"1999"},{"issue":"4","key":"10.1016\/j.neunet.2026.108841_bib0122","doi-asserted-by":"crossref","first-page":"561","DOI":"10.1016\/j.jco.2003.11.004","article-title":"On the tractability of multivariate integration and approximation by neural networks","volume":"20","author":"Mhaskar","year":"2004","journal-title":"Journal of Complexity"},{"issue":"7","key":"10.1016\/j.neunet.2026.108841_bib0123","doi-asserted-by":"crossref","first-page":"989","DOI":"10.1016\/j.neunet.2004.04.001","article-title":"When is approximation by Gaussian networks necessarily a linear process?","volume":"17","author":"Mhaskar","year":"2004","journal-title":"Neural Networks"},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0124","doi-asserted-by":"crossref","first-page":"215","DOI":"10.1016\/j.acha.2004.11.004","article-title":"On the representation of smooth functions on the sphere using finitely many bits","volume":"18","author":"Mhaskar","year":"2005","journal-title":"Applied and Computational Harmonic Analysis"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0125","doi-asserted-by":"crossref","first-page":"63","DOI":"10.1016\/j.acha.2009.08.006","article-title":"Eignets for function approximation on manifolds","volume":"29","author":"Mhaskar","year":"2010","journal-title":"Applied and Computational Harmonic Analysis"},{"issue":"4","key":"10.1016\/j.neunet.2026.108841_bib0126","doi-asserted-by":"crossref","first-page":"345","DOI":"10.1016\/j.neunet.2010.12.007","article-title":"A generalized diffusion frame for parsimonious representation of functions on data defined manifolds","volume":"24","author":"Mhaskar","year":"2011","journal-title":"Neural Networks"},{"key":"10.1016\/j.neunet.2026.108841_bib0127","series-title":"Festschrift for the 80th birthday of Ian sloan","first-page":"931","article-title":"Approximate quadrature measures on data\u2013defined spaces","author":"Mhaskar","year":"2017"},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0128","doi-asserted-by":"crossref","first-page":"611","DOI":"10.1016\/j.acha.2016.06.007","article-title":"A unified framework for harmonic analysis of functions on directed graphs and changing data","volume":"44","author":"Mhaskar","year":"2018","journal-title":"Applied and Computational Harmonic Analysis"},{"key":"10.1016\/j.neunet.2026.108841_bib0129","unstructured":"Mhaskar, H. N. (2019). Dimension independent bounds for general shallow networks. arXiv preprint arXiv: 1908.09880."},{"key":"10.1016\/j.neunet.2026.108841_bib0130","doi-asserted-by":"crossref","first-page":"142","DOI":"10.1016\/j.neunet.2019.11.006","article-title":"Dimension independent bounds for general shallow networks","volume":"123","author":"Mhaskar","year":"2020","journal-title":"Neural Networks"},{"key":"10.1016\/j.neunet.2026.108841_bib0131","doi-asserted-by":"crossref","first-page":"253","DOI":"10.1016\/j.neunet.2020.08.018","article-title":"A direct method for function approximation on data defined manifolds","volume":"132","author":"Mhaskar","year":"2020","journal-title":"Neural Networks"},{"key":"10.1016\/j.neunet.2026.108841_bib0132","doi-asserted-by":"crossref","first-page":"30","DOI":"10.3389\/fams.2020.00030","article-title":"Kernel-based analysis of massive data","volume":"6","author":"Mhaskar","year":"2020","journal-title":"Frontiers in Applied Mathematics and Statistics"},{"key":"10.1016\/j.neunet.2026.108841_bib0133","unstructured":"Mhaskar, H. N. (2022). Local approximation of operators. arXiv preprint arXiv: 2202.06392."},{"key":"10.1016\/j.neunet.2026.108841_bib0134","doi-asserted-by":"crossref","DOI":"10.1016\/j.acha.2023.01.004","article-title":"Local approximation of operators","author":"Mhaskar","year":"2023","journal-title":"Applied and Computational Harmonic Analysis"},{"key":"10.1016\/j.neunet.2026.108841_bib0135","doi-asserted-by":"crossref","DOI":"10.1016\/j.neunet.2025.107282","article-title":"Approximation by non-symmetric networks for cross-domain learning","author":"Mhaskar","year":"2025","journal-title":"Neural Networks"},{"key":"10.1016\/j.neunet.2026.108841_bib0136","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1016\/j.jco.2018.09.002","article-title":"Function approximation with zonal function networks with activation functions analogous to the rectified linear unit functions","volume":"51","author":"Mhaskar","year":"2019","journal-title":"Journal of Complexity"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0137","doi-asserted-by":"crossref","first-page":"143","DOI":"10.1162\/neco.1997.9.1.143","article-title":"Neural networks for functional approximation and system identification","volume":"9","author":"Mhaskar","year":"1997","journal-title":"Neural Computation"},{"key":"10.1016\/j.neunet.2026.108841_bib0138","doi-asserted-by":"crossref","unstructured":"Mhaskar, H. N., Kitimoon, S., & Raj, R. G. (2024). Robust and tractable multidimensional exponential analysis. arXiv preprint arXiv: 2404.11004.","DOI":"10.1007\/s11075-025-02259-7"},{"issue":"03","key":"10.1016\/j.neunet.2026.108841_bib0139","doi-asserted-by":"crossref","first-page":"535","DOI":"10.1142\/S0219530524400013","article-title":"Tractability of approximation by general shallow networks","volume":"22","author":"Mhaskar","year":"2024","journal-title":"Analysis and Applications"},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0140","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"},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0141","doi-asserted-by":"crossref","first-page":"277","DOI":"10.1147\/rd.383.0277","article-title":"Dimension-independent bounds on the degree of approximation by neural networks","volume":"38","author":"Mhaskar","year":"1994","journal-title":"IBM Journal of Research and Development"},{"issue":"2","key":"10.1016\/j.neunet.2026.108841_bib0142","doi-asserted-by":"crossref","first-page":"151","DOI":"10.1006\/aama.1995.1008","article-title":"Degree of approximation by neural and translation networks with a single hidden layer","volume":"16","author":"Mhaskar","year":"1995","journal-title":"Advances in Applied Mathematics"},{"issue":"5","key":"10.1016\/j.neunet.2026.108841_bib0143","doi-asserted-by":"crossref","first-page":"1355","DOI":"10.1090\/S0002-9939-01-06240-2","article-title":"Approximation with interpolatory constraints","volume":"130","author":"Mhaskar","year":"2002","journal-title":"Proceedings of the American Mathematical Society"},{"issue":"235","key":"10.1016\/j.neunet.2026.108841_bib0144","doi-asserted-by":"crossref","first-page":"1113","DOI":"10.1090\/S0025-5718-00-01240-0","article-title":"Spherical Marcinkiewicz-Zygmund inequalities and positive quadrature","volume":"70","author":"Mhaskar","year":"2001","journal-title":"Mathematics of Computation"},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0145","doi-asserted-by":"crossref","first-page":"485","DOI":"10.1007\/s13373-013-0043-1","article-title":"Applications of classical approximation theory to periodic basis function networks and computational harmonic analysis","volume":"3","author":"Mhaskar","year":"2013","journal-title":"Bulletin of Mathematical Sciences"},{"key":"10.1016\/j.neunet.2026.108841_bib0146","unstructured":"Mhaskar, H. N., & O\u2019Dowd, R. (2025). Signal separation approach for classification. In preparation."},{"key":"10.1016\/j.neunet.2026.108841_bib0147","unstructured":"Mhaskar, H. N., & O\u2019Dowd, R. (2023). Local transfer learning from one data space to another. arXiv preprint arXiv: 2302.00160."},{"key":"10.1016\/j.neunet.2026.108841_bib0148","doi-asserted-by":"crossref","unstructured":"Mhaskar, H. N., O\u2019Dowd, R., & Tsoukanis, E. (2025). Active learning classification from a signal separation perspective. arXiv preprint arXiv: 2502.16425.","DOI":"10.1109\/SampTA64769.2025.11133508"},{"key":"10.1016\/j.neunet.2026.108841_bib0149","doi-asserted-by":"crossref","DOI":"10.1016\/j.neunet.2024.106759","article-title":"Learning on manifolds without manifold learning","volume":"181","author":"Mhaskar","year":"2025","journal-title":"Neural Networks"},{"key":"10.1016\/j.neunet.2026.108841_bib0150","unstructured":"Mhaskar, H. N., Pereverzyev, S. V., Semenov, V. Y., & Semenova, E. V. (2018). Data based construction of kernels for classification. Submitted for publication, https:\/\/www.ricam.oeaw.ac.at\/files\/reports\/18\/rep18-25.pdf."},{"key":"10.1016\/j.neunet.2026.108841_bib0151","doi-asserted-by":"crossref","first-page":"14","DOI":"10.3389\/fams.2017.00014","article-title":"A deep learning approach to diabetic blood glucose prediction","volume":"3","author":"Mhaskar","year":"2017","journal-title":"Frontiers in Applied Mathematics and Statistics"},{"issue":"06","key":"10.1016\/j.neunet.2026.108841_bib0152","doi-asserted-by":"crossref","first-page":"829","DOI":"10.1142\/S0219530516400042","article-title":"Deep vs. shallow networks: An approximation theory perspective","volume":"14","author":"Mhaskar","year":"2016","journal-title":"Analysis and Applications"},{"key":"10.1016\/j.neunet.2026.108841_bib0153","doi-asserted-by":"crossref","unstructured":"Mhaskar, H. N., & Poggio, T. (2018). An analysis of training and generalization errors in shallow and deep networks. 121, 229\u2013241. arXiv preprint arXiv:1802.06266.","DOI":"10.1016\/j.neunet.2019.08.028"},{"issue":"2-3","key":"10.1016\/j.neunet.2026.108841_bib0154","doi-asserted-by":"crossref","first-page":"95","DOI":"10.1023\/A:1018921319865","article-title":"On the detection of singularities of a periodic function","volume":"12","author":"Mhaskar","year":"2000","journal-title":"Advances in Computational Mathematics"},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0155","doi-asserted-by":"crossref","first-page":"353","DOI":"10.1007\/s00041-005-4006-0","article-title":"On local smoothness classes of periodic functions","volume":"11","author":"Mhaskar","year":"2005","journal-title":"Journal of Fourier Analysis and Applications"},{"key":"10.1016\/j.neunet.2026.108841_bib0156","series-title":"Interpolation of scattered data: Distance matrices and conditionally positive definite functions","author":"Micchelli","year":"1984"},{"key":"10.1016\/j.neunet.2026.108841_bib0157","series-title":"Lectures on constructive approximation: Fourier, spline, and wavelet methods on the real line, the sphere, and the ball","author":"Michel","year":"2012"},{"key":"10.1016\/j.neunet.2026.108841_bib0158","doi-asserted-by":"crossref","DOI":"10.1016\/j.jqsrt.2021.107705","article-title":"Physics informed neural networks for simulating radiative transfer","volume":"270","author":"Mishra","year":"2021","journal-title":"Journal of Quantitative Spectroscopy and Radiative Transfer"},{"issue":"2","key":"10.1016\/j.neunet.2026.108841_bib0159","doi-asserted-by":"crossref","first-page":"981","DOI":"10.1093\/imanum\/drab032","article-title":"Estimates on the generalization error of physics-informed neural networks for approximating a class of inverse problems for PDEs","volume":"42","author":"Mishra","year":"2022","journal-title":"IMA Journal of Numerical Analysis"},{"key":"10.1016\/j.neunet.2026.108841_bib0160","doi-asserted-by":"crossref","first-page":"161","DOI":"10.1007\/s10444-004-7634-z","article-title":"Learning theory: Stability is sufficient for generalization and necessary and sufficient for consistency of empirical risk minimization","volume":"25","author":"Mukherjee","year":"2006","journal-title":"Advances in Computational Mathematics"},{"key":"10.1016\/j.neunet.2026.108841_bib0161","series-title":"Probabilistic machine learning: an introduction","author":"Murphy","year":"2022"},{"issue":"2","key":"10.1016\/j.neunet.2026.108841_bib0162","doi-asserted-by":"crossref","first-page":"246","DOI":"10.1162\/neco.1991.3.2.246","article-title":"Universal approximation using radial-basis-function networks","volume":"3","author":"Park","year":"1991","journal-title":"Neural Computation"},{"key":"10.1016\/j.neunet.2026.108841_bib0163","doi-asserted-by":"crossref","DOI":"10.1016\/j.jcp.2023.112464","article-title":"A unified scalable framework for causal sweeping strategies for physics-informed neural networks (PINNs) and their temporal decompositions","volume":"493","author":"Penwarden","year":"2023","journal-title":"Journal of Computational Physics"},{"key":"10.1016\/j.neunet.2026.108841_bib0164","unstructured":"Perekrestenko, D., Grohs, P., Elbr\u00e4chter, D., & B\u00f6lcskei, H. (2018). The universal approximation power of finite-width deep reLU networks. arXiv preprint arXiv: 1806.01528."},{"key":"10.1016\/j.neunet.2026.108841_bib0165","doi-asserted-by":"crossref","first-page":"296","DOI":"10.1016\/j.neunet.2018.08.019","article-title":"Optimal approximation of piecewise smooth functions using deep reLU neural networks","volume":"108","author":"Petersen","year":"2018","journal-title":"Neural Networks"},{"issue":"2","key":"10.1016\/j.neunet.2026.108841_bib0166","doi-asserted-by":"crossref","first-page":"759","DOI":"10.1007\/s00365-022-09576-3","article-title":"Lipschitz widths","volume":"57","author":"Petrova","year":"2023","journal-title":"Constructive Approximation"},{"key":"10.1016\/j.neunet.2026.108841_bib0167","doi-asserted-by":"crossref","DOI":"10.1016\/j.cma.2024.116996","article-title":"RiemannONets: Interpretable neural operators for riemann problems","volume":"426","author":"Peyvan","year":"2024","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"key":"10.1016\/j.neunet.2026.108841_bib0168","doi-asserted-by":"crossref","first-page":"143","DOI":"10.1017\/S0962492900002919","article-title":"Approximation theory of the MLP model in neural networks","volume":"8","author":"Pinkus","year":"1999","journal-title":"Acta Numerica"},{"key":"10.1016\/j.neunet.2026.108841_bib0169","series-title":"N-widths in approximation theory","volume":"vol. 7","author":"Pinkus","year":"2012"},{"key":"10.1016\/j.neunet.2026.108841_bib0170","first-page":"1","article-title":"Why and when can deep-but not shallow-networks avoid the curse of dimensionality: A review","author":"Poggio","year":"2017","journal-title":"International Journal of Automation and Computing"},{"key":"10.1016\/j.neunet.2026.108841_bib0171","doi-asserted-by":"crossref","first-page":"353","DOI":"10.1016\/S0024-3795(02)00592-X","article-title":"Fast algorithms for discrete polynomial transforms on arbitrary grids","volume":"366","author":"Potts","year":"2003","journal-title":"Linear Algebra and its Applications"},{"issue":"224","key":"10.1016\/j.neunet.2026.108841_bib0172","doi-asserted-by":"crossref","first-page":"1577","DOI":"10.1090\/S0025-5718-98-00975-2","article-title":"Fast algorithms for discrete polynomial transforms","volume":"67","author":"Potts","year":"1998","journal-title":"Mathematics of Computation"},{"issue":"6690","key":"10.1016\/j.neunet.2026.108841_bib0173","doi-asserted-by":"crossref","first-page":"1461","DOI":"10.1126\/science.adi5639","article-title":"Mechanism for feature learning in neural networks and backpropagation-free machine learning models","volume":"383","author":"Radhakrishnan","year":"2024","journal-title":"Science"},{"key":"10.1016\/j.neunet.2026.108841_bib0174","doi-asserted-by":"crossref","unstructured":"Radhakrishnan, A., Belkin, M., & Drusvyatskiy, D. (2024b). Linear recursive feature machines provably recover low-rank matrices. arXiv preprint arXiv: 2401.04553.","DOI":"10.1073\/pnas.2411325122"},{"key":"10.1016\/j.neunet.2026.108841_bib0175","series-title":"International conference on machine learning","first-page":"5301","article-title":"On the spectral bias of neural networks","author":"Rahaman","year":"2019"},{"key":"10.1016\/j.neunet.2026.108841_bib0176","doi-asserted-by":"crossref","first-page":"686","DOI":"10.1016\/j.jcp.2018.10.045","article-title":"Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations","volume":"378","author":"Raissi","year":"2019","journal-title":"Journal of Computational Physics"},{"key":"10.1016\/j.neunet.2026.108841_bib0177","series-title":"ICLR 2023 workshop on physics for machine learning","article-title":"Convolutional neural operators","author":"Raonic","year":"2023"},{"key":"10.1016\/j.neunet.2026.108841_bib0178","series-title":"Chebyshev polynomials","author":"Rivlin","year":"2020"},{"key":"10.1016\/j.neunet.2026.108841_bib0179","unstructured":"Schmidt-Hieber, J. (2019). Deep ReLU network approximation of functions on a manifold. arXiv preprint arXiv: 1908.00695."},{"issue":"4","key":"10.1016\/j.neunet.2026.108841_bib0180","first-page":"1875","article-title":"Nonparametric regression using deep neural networks with ReLU activation function","volume":"48","author":"Schmidt-Hieber","year":"2020","journal-title":"The Annals of Statistics"},{"key":"10.1016\/j.neunet.2026.108841_bib0181","doi-asserted-by":"crossref","first-page":"119","DOI":"10.1016\/j.neunet.2021.01.020","article-title":"The Kolmogorov\u2013Arnold representation theorem revisited","volume":"137","author":"Schmidt-Hieber","year":"2021","journal-title":"Neural Networks"},{"issue":"3","key":"10.1016\/j.neunet.2026.108841_bib0182","doi-asserted-by":"crossref","first-page":"537","DOI":"10.1016\/j.acha.2016.04.003","article-title":"Provable approximation properties for deep neural networks","volume":"44","author":"Shaham","year":"2018","journal-title":"Applied and Computational Harmonic Analysis"},{"key":"10.1016\/j.neunet.2026.108841_bib0183","doi-asserted-by":"crossref","unstructured":"Shin, Y., Darbon, J., & Karniadakis, G. E. (2020a). On the convergence of physics informed neural networks for linear second-order elliptic and parabolic type PDEs. arXiv preprint arXiv: 2004.01806.","DOI":"10.4208\/cicp.OA-2020-0193"},{"key":"10.1016\/j.neunet.2026.108841_bib0184","unstructured":"Shin, Y., Zhang, Z., & Karniadakis, G. E. (2020b). Error estimates of residual minimization using neural networks for linear PDEs. arXiv preprint arXiv: 2010.08019."},{"key":"10.1016\/j.neunet.2026.108841_bib0185","first-page":"293","article-title":"Theoretical foundations of physics-informed neural networks and deep neural operators: A brief review","volume":"25","author":"Shin","year":"2024","journal-title":"Handbook of Numerical Analysis"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0186","doi-asserted-by":"crossref","first-page":"68","DOI":"10.1109\/MSP.2021.3118904","article-title":"A physics-informed neural network for quantifying the microstructural properties of polycrystalline nickel using ultrasound data: A promising approach for solving inverse problems","volume":"39","author":"Shukla","year":"2021","journal-title":"IEEE Signal Processing Magazine"},{"key":"10.1016\/j.neunet.2026.108841_bib0187","doi-asserted-by":"crossref","DOI":"10.1016\/j.jcp.2021.110683","article-title":"Parallel physics-informed neural networks via domain decomposition","volume":"447","author":"Shukla","year":"2021","journal-title":"Journal of Computational Physics"},{"issue":"357","key":"10.1016\/j.neunet.2026.108841_bib0188","first-page":"1","article-title":"Optimal approximation rates for deep ReLU neural networks on Sobolev and Besov spaces","volume":"24","author":"Siegel","year":"2023","journal-title":"Journal of Machine Learning Research"},{"key":"10.1016\/j.neunet.2026.108841_bib0189","doi-asserted-by":"crossref","DOI":"10.1016\/j.jco.2024.101884","article-title":"Sharp lower bounds on the manifold widths of Sobolev and Besov spaces","author":"Siegel","year":"2024","journal-title":"Journal of Complexity"},{"key":"10.1016\/j.neunet.2026.108841_bib0190","first-page":"1","article-title":"Optimal approximation of zonoids and uniform approximation by shallow neural networks","author":"Siegel","year":"2025","journal-title":"Constructive Approximation"},{"key":"10.1016\/j.neunet.2026.108841_bib0191","first-page":"1","article-title":"Sharp bounds on the approximation rates, metric entropy, and n-widths of shallow neural networks","author":"Siegel","year":"2022","journal-title":"Foundations of Computational Mathematics"},{"issue":"1","key":"10.1016\/j.neunet.2026.108841_bib0192","doi-asserted-by":"crossref","first-page":"128","DOI":"10.1016\/j.acha.2006.03.004","article-title":"From graph to manifold laplacian: The convergence rate","volume":"21","author":"Singer","year":"2006","journal-title":"Applied and Computational Harmonic Analysis"},{"key":"10.1016\/j.neunet.2026.108841_bib0193","series-title":"Best approximation in normed linear spaces by elements of linear subspaces","author":"Singer","year":"1970"},{"key":"10.1016\/j.neunet.2026.108841_bib0194","doi-asserted-by":"crossref","first-page":"424","DOI":"10.1016\/j.neunet.2023.07.012","article-title":"Approximation of smooth functionals using deep ReLU networks","volume":"166","author":"Song","year":"2023","journal-title":"Neural Networks"},{"key":"10.1016\/j.neunet.2026.108841_bib0195","series-title":"Singular integrals and differentiability properties of functions (PMS-30)","volume":"vol. 30","author":"Stein","year":"2016"},{"key":"10.1016\/j.neunet.2026.108841_bib0196","unstructured":"Sule, S., Evans, L., & Cameron, M. (2023). Sharp error estimates for target measure diffusion maps with applications to the committor problem. arXiv preprint arXiv: 2312.14418."},{"key":"10.1016\/j.neunet.2026.108841_bib0197","first-page":"805","article-title":"On some convergent interpolatory polynomials","volume":"19","author":"Szabados","year":"1978","journal-title":"Fourier Analysis and Approximation Theory, Coll. Math. Soc. J\u00e1nos Bolyai"},{"key":"10.1016\/j.neunet.2026.108841_bib0198","series-title":"Theory of approximation of functions of a real variable: International series of monographs on pure and applied mathematics","volume":"vol. 34","author":"Timan","year":"2014"},{"key":"10.1016\/j.neunet.2026.108841_bib0199","series-title":"Fourier analysis and approximation of functions","author":"Trigub","year":"2012"},{"key":"10.1016\/j.neunet.2026.108841_bib0200","doi-asserted-by":"crossref","DOI":"10.1016\/j.cma.2022.115783","article-title":"Wavelet neural operator for solving parametric partial differential equations in computational mechanics problems","volume":"404","author":"Tripura","year":"2023","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"key":"10.1016\/j.neunet.2026.108841_bib0201","series-title":"The nature of statistical learning theory","author":"Vapnik","year":"2013"},{"issue":"2","key":"10.1016\/j.neunet.2026.108841_bib0202","doi-asserted-by":"crossref","first-page":"431","DOI":"10.1002\/cpa.3160200210","article-title":"On the behavior of the fundamental solution of the heat equation with variable coefficients","volume":"20","author":"Varadhan","year":"1967","journal-title":"Communications on Pure and Applied Mathematics"},{"key":"10.1016\/j.neunet.2026.108841_bib0203","first-page":"555","article-title":"Consistency of spectral clustering","author":"Von Luxburg","year":"2008","journal-title":"The Annals of Statistics"},{"issue":"5","key":"10.1016\/j.neunet.2026.108841_bib0204","doi-asserted-by":"crossref","first-page":"A3055","DOI":"10.1137\/20M1318043","article-title":"Understanding and mitigating gradient pathologies in physics-informed neural networks","volume":"43","author":"Wang","year":"2021","journal-title":"SIAM Journal on Scientific Computing"},{"key":"10.1016\/j.neunet.2026.108841_bib0205","series-title":"Algebraic geometry and statistical learning theory","volume":"vol. 25","author":"Watanabe","year":"2009"},{"key":"10.1016\/j.neunet.2026.108841_bib0206","unstructured":"Xu, J. (2020). The finite neuron method and convergence analysis. arXiv preprint arXiv: 2010.01458."},{"key":"10.1016\/j.neunet.2026.108841_bib0207","first-page":"1","article-title":"Overview frequency principle\/spectral bias in deep learning","author":"Xu","year":"2024","journal-title":"Communications on Applied Mathematics and Computation"},{"key":"10.1016\/j.neunet.2026.108841_bib0208","doi-asserted-by":"crossref","unstructured":"Yarotsky, D. (2016). Error bounds for approximations with deep ReLU networks. arXiv preprint arXiv: 1610.01145, 94, 103\u2013114.","DOI":"10.1016\/j.neunet.2017.07.002"},{"key":"10.1016\/j.neunet.2026.108841_bib0209","unstructured":"Yun, C., Bhojanapalli, S., Rawat, A. S., Reddi, S. J., & Kumar, S. (2019). Are transformers universal approximators of sequence-to-sequence functions?arXiv preprint arXiv: 1912.10077."},{"key":"10.1016\/j.neunet.2026.108841_bib0210","article-title":"A unified framework for the error analysis of physics-informed neural networks","author":"Zeinhofer","year":"2024","journal-title":"IMA Journal of Numerical Analysis"},{"key":"10.1016\/j.neunet.2026.108841_bib0211","doi-asserted-by":"crossref","DOI":"10.1016\/j.cma.2025.118667","article-title":"BubbleOKAN: A physics-informed interpretable neural operator for high-frequency bubble dynamics","volume":"450","author":"Zhang","year":"2026","journal-title":"Computer Methods in Applied Mechanics and Engineering"},{"issue":"2","key":"10.1016\/j.neunet.2026.108841_bib0212","doi-asserted-by":"crossref","first-page":"787","DOI":"10.1016\/j.acha.2019.06.004","article-title":"Universality of deep convolutional neural networks","volume":"48","author":"Zhou","year":"2020","journal-title":"Applied and Computational Harmonic Analysis"}],"container-title":["Neural Networks"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/api.elsevier.com\/content\/article\/PII:S0893608026003035?httpAccept=text\/xml","content-type":"text\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/api.elsevier.com\/content\/article\/PII:S0893608026003035?httpAccept=text\/plain","content-type":"text\/plain","content-version":"vor","intended-application":"text-mining"}],"deposited":{"date-parts":[[2026,8,13]],"date-time":"2026-08-13T15:13:25Z","timestamp":1786634005000},"score":1,"resource":{"primary":{"URL":"https:\/\/linkinghub.elsevier.com\/retrieve\/pii\/S0893608026003035"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8]]},"references-count":212,"alternative-id":["S0893608026003035"],"URL":"https:\/\/doi.org\/10.1016\/j.neunet.2026.108841","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":"An approximation theory perspective on machine learning","name":"articletitle","label":"Article Title"},{"value":"Neural Networks","name":"journaltitle","label":"Journal Title"},{"value":"https:\/\/doi.org\/10.1016\/j.neunet.2026.108841","name":"articlelink","label":"CrossRef DOI link to publisher maintained version"},{"value":"article","name":"content_type","label":"Content Type"},{"value":"\u00a9 2026 The Author(s). Published by Elsevier Ltd.","name":"copyright","label":"Copyright"}],"article-number":"108841"}}