{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,12,24]],"date-time":"2025-12-24T12:24:31Z","timestamp":1766579071458,"version":"3.37.3"},"reference-count":49,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"3","license":[{"start":{"date-parts":[[2025,3,1]],"date-time":"2025-03-01T00:00:00Z","timestamp":1740787200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2025,3,1]],"date-time":"2025-03-01T00:00:00Z","timestamp":1740787200000},"content-version":"am","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2025,3,1]],"date-time":"2025-03-01T00:00:00Z","timestamp":1740787200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,3,1]],"date-time":"2025-03-01T00:00:00Z","timestamp":1740787200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"name":"National Science Foundation (NSF) CAREER Award","award":["CCF-2236829"],"award-info":[{"award-number":["CCF-2236829"]}]},{"name":"U.S. Army Research Office Early Career Award","award":["W911NF-21-1-0242"],"award-info":[{"award-number":["W911NF-21-1-0242"]}]},{"DOI":"10.13039\/100000006","name":"Office of Naval Research","doi-asserted-by":"publisher","award":["N00014-24-1-2164"],"award-info":[{"award-number":["N00014-24-1-2164"]}],"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":["N00014-20-1-2157"],"award-info":[{"award-number":["N00014-20-1-2157"]}],"id":[{"id":"10.13039\/100000006","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000001","name":"National Science Foundation","doi-asserted-by":"publisher","award":["DMS-2032014"],"award-info":[{"award-number":["DMS-2032014"]}],"id":[{"id":"10.13039\/100000001","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000893","name":"Simons Foundation","doi-asserted-by":"publisher","award":["814641"],"award-info":[{"award-number":["814641"]}],"id":[{"id":"10.13039\/100000893","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000001","name":"National Science Foundation","doi-asserted-by":"publisher","award":["DMS-2134248"],"award-info":[{"award-number":["DMS-2134248"]}],"id":[{"id":"10.13039\/100000001","id-type":"DOI","asserted-by":"publisher"}]},{"name":"NSF CAREER Award","award":["CCF-2236829"],"award-info":[{"award-number":["CCF-2236829"]}]},{"name":"U.S. Army Research Office Early Career Award","award":["W911NF-21-1-0242"],"award-info":[{"award-number":["W911NF-21-1-0242"]}]},{"DOI":"10.13039\/100000006","name":"Office of Naval Research","doi-asserted-by":"publisher","award":["N00014-24-1-2164"],"award-info":[{"award-number":["N00014-24-1-2164"]}],"id":[{"id":"10.13039\/100000006","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Inform. Theory"],"published-print":{"date-parts":[[2025,3]]},"DOI":"10.1109\/tit.2025.3530355","type":"journal-article","created":{"date-parts":[[2025,1,15]],"date-time":"2025-01-15T19:40:15Z","timestamp":1736970015000},"page":"1926-1977","source":"Crossref","is-referenced-by-count":2,"title":["Overparameterized ReLU Neural Networks Learn the Simplest Model: Neural Isometry and Phase Transitions"],"prefix":"10.1109","volume":"71","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-0364-0893","authenticated-orcid":false,"given":"Yifei","family":"Wang","sequence":"first","affiliation":[{"name":"Department of Electrical Engineering, Stanford University, Stanford, CA, USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0975-352X","authenticated-orcid":false,"given":"Yixuan","family":"Hua","sequence":"additional","affiliation":[{"name":"Department of Operations Research and Financial Engineering, Princeton University, Princeton, NJ, USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9234-924X","authenticated-orcid":false,"given":"Emmanuel J.","family":"Cand\u00e8s","sequence":"additional","affiliation":[{"name":"Department of Statistics, Department of Mathematics, Stanford University, Stanford, CA, USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-0870-9992","authenticated-orcid":false,"given":"Mert","family":"Pilanci","sequence":"additional","affiliation":[{"name":"Department of Electrical Engineering, Stanford University, Stanford, CA, USA"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"doi-asserted-by":"publisher","key":"ref1","DOI":"10.1093\/imaiai\/iau005"},{"key":"ref2","first-page":"7696","article-title":"Neural networks are convex regularizers: Exact polynomial-time convex optimization formulations for two-layer networks","volume-title":"Proc. Int. Conf. Mach. Learn. (ICML)","author":"Pilanci"},{"issue":"212","key":"ref3","first-page":"1","article-title":"Convex geometry and duality of over-parameterized neural networks","volume":"22","author":"Ergen","year":"2021","journal-title":"J. Mach. Learn. Res."},{"key":"ref4","first-page":"2993","article-title":"Global optimality beyond two layers: Training deep ReLU networks via convex programs","volume-title":"Proc. Int. Conf. Mach. Learn. (ICLR)","author":"Ergen"},{"volume-title":"Proc. Int. Conf. Learn. Represent. (ICLR)","author":"Wang","article-title":"The convex geometry of backpropagation: Neural network gradient flows converge to extreme points of the dual convex program","key":"ref5"},{"volume-title":"Proc. Int. Conf. Learn. Represent. (ICLR)","author":"Wang","article-title":"The hidden convex optimization landscape of two-layer ReLU neural networks: An exact characterization of the optimal solutions","key":"ref6"},{"key":"ref7","first-page":"2541","article-title":"On model selection consistency of lasso","volume":"7","author":"Zhao","year":"2006","journal-title":"J. Mach. Learn. Res."},{"doi-asserted-by":"publisher","key":"ref8","DOI":"10.1109\/TIT.2005.858979"},{"doi-asserted-by":"publisher","key":"ref9","DOI":"10.1016\/j.crma.2008.03.014"},{"key":"ref10","first-page":"2256","article-title":"Deep unsupervised learning using nonequilibrium thermodynamics","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Sohl-Dickstein"},{"key":"ref11","first-page":"6840","article-title":"Denoising diffusion probabilistic models","volume-title":"Proc. NIPS","volume":"33","author":"Ho"},{"volume-title":"Proc. Int. Conf. Learn. Represent. (ICLR)","author":"Ergen","article-title":"Demystifying batch normalization in ReLU networks: Equivalent convex optimization models and implicit regularization","key":"ref12"},{"volume-title":"Proc. Int. Conf. Learn. Represent. (ICLR)","author":"Ergen","article-title":"Implicit convex regularizers of CNN architectures: Convex optimization of two- and three-layer networks in polynomial time","key":"ref13"},{"doi-asserted-by":"publisher","key":"ref14","DOI":"10.1007\/s10107-024-02153-5"},{"key":"ref15","first-page":"19050","article-title":"Unraveling attention via convex duality: Analysis and interpretations of vision transformers","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Sahiner"},{"volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Sahiner","article-title":"Hidden convexity of Wasserstein GANs: Interpretable generative models with closed-form solutions","key":"ref16"},{"doi-asserted-by":"publisher","key":"ref17","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref18","first-page":"1","article-title":"Neural tangent kernel: Convergence and generalization in neural networks","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"31","author":"Jacot"},{"volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Du","article-title":"Gradient descent provably optimizes over-parameterized neural networks","key":"ref19"},{"key":"ref20","first-page":"15954","article-title":"On the linearity of large non-linear models: When and why the tangent kernel is constant","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"33","author":"Liu"},{"doi-asserted-by":"publisher","key":"ref21","DOI":"10.1007\/s10994-019-05839-6"},{"key":"ref22","first-page":"242","article-title":"A convergence theory for deep learning via over-parameterization","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Allen-Zhu"},{"key":"ref23","first-page":"6155","article-title":"Learning and generalization in overparameterized neural networks, going beyond two layers","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"32","author":"Allen-Zhu"},{"key":"ref24","first-page":"1","article-title":"On lazy training in differentiable programming","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"32","author":"Chizat"},{"key":"ref25","first-page":"322","article-title":"Fine-grained analysis of optimization and generalization for overparameterized two-layer neural networks","volume-title":"Proc. Int. Conf. Mach. Learn. (ICML)","author":"Arora"},{"volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Chen","article-title":"How much over-parameterization is sufficient to learn deep ReLU networks?","key":"ref26"},{"volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Neyshabur","article-title":"The role of over-parametrization in generalization of neural networks","key":"ref27"},{"doi-asserted-by":"publisher","key":"ref28","DOI":"10.1162\/neco.1997.9.1.1"},{"key":"ref29","article-title":"Three factors influencing minima in SGD","author":"Jastrz\u0229bski","year":"2017","journal-title":"arXiv:1711.04623"},{"key":"ref30","article-title":"Towards understanding generalization of deep learning: Perspective of loss landscapes","author":"Wu","year":"2017","journal-title":"arXiv:1706.10239"},{"key":"ref31","article-title":"Accurate, large minibatch SGD: Training ImageNet in 1 hour","author":"Goyal","year":"2017","journal-title":"arXiv:1706.02677"},{"key":"ref32","first-page":"1","article-title":"Train longer, generalize better: Closing the generalization gap in large batch training of neural networks","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"30","author":"Hoffer"},{"key":"ref33","article-title":"How many factors influence minima in SGD?","author":"Luo","year":"2020","journal-title":"arXiv:2009.11858"},{"key":"ref34","first-page":"1","article-title":"Learning two-layer neural networks with symmetric inputs","volume-title":"Proc. Int. Conf. Learn. Represent.","author":"Ge"},{"key":"ref35","first-page":"195","article-title":"Learning two layer rectified neural networks in polynomial time","volume-title":"Proc. Conf. Learn. Theory","author":"Bakshi"},{"key":"ref36","first-page":"3404","article-title":"An analytical formula of population gradient for two-layered ReLU network and its applications in convergence and critical point analysis","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Tian"},{"key":"ref37","first-page":"2007","article-title":"Learning ReLUs via gradient descent","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"30","author":"Soltanolkotabi"},{"key":"ref38","first-page":"597","article-title":"Convergence analysis of two-layer neural networks with ReLU activation","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"30","author":"Li"},{"key":"ref39","first-page":"1524","article-title":"Learning one-hidden-layer ReLU networks via gradient descent","volume-title":"Proc. 22nd Int. Conf. Artif. Intell. Statist.","author":"Zhang"},{"key":"ref40","first-page":"4140","article-title":"Recovery guarantees for one-hidden-layer neural networks","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Zhong"},{"doi-asserted-by":"publisher","key":"ref41","DOI":"10.1109\/PGEC.1965.264137"},{"volume-title":"Proc. 41st Int. Conf. Mach. Learn.","author":"Kim","article-title":"Convex relaxations of ReLU neural networks approximate global optima in polynomial time","key":"ref42"},{"key":"ref43","first-page":"15770","article-title":"Fast convex optimization for two-layer ReLU networks: Equivalent model classes and cone decompositions","volume-title":"Proc. Int. Conf. Mach. Learn. (ICML)","author":"Mishkin"},{"doi-asserted-by":"publisher","key":"ref44","DOI":"10.1111\/j.1467-9868.2005.00532.x"},{"doi-asserted-by":"publisher","key":"ref45","DOI":"10.1017\/9781108627771"},{"doi-asserted-by":"publisher","key":"ref46","DOI":"10.1017\/9781108231596"},{"doi-asserted-by":"publisher","key":"ref47","DOI":"10.1214\/09-EJS506"},{"volume-title":"Decoupling: From Dependence to Independence","year":"2012","author":"De la Pena","key":"ref48"},{"doi-asserted-by":"publisher","key":"ref49","DOI":"10.1214\/20-AOS1990"}],"container-title":["IEEE Transactions on Information Theory"],"original-title":[],"link":[{"URL":"https:\/\/ieeexplore.ieee.org\/ielam\/18\/10896902\/10843278-aam.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/18\/10896902\/10843278.pdf?arnumber=10843278","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,2,21]],"date-time":"2025-02-21T18:40:29Z","timestamp":1740163229000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/10843278\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,3]]},"references-count":49,"journal-issue":{"issue":"3"},"URL":"https:\/\/doi.org\/10.1109\/tit.2025.3530355","relation":{},"ISSN":["0018-9448","1557-9654"],"issn-type":[{"type":"print","value":"0018-9448"},{"type":"electronic","value":"1557-9654"}],"subject":[],"published":{"date-parts":[[2025,3]]}}}