{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,3]],"date-time":"2026-07-03T03:16:43Z","timestamp":1783048603016,"version":"3.54.6"},"reference-count":52,"publisher":"Elsevier BV","license":[{"start":{"date-parts":[[2026,9,1]],"date-time":"2026-09-01T00:00:00Z","timestamp":1788220800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/www.elsevier.com\/tdm\/userlicense\/1.0\/"},{"start":{"date-parts":[[2026,9,1]],"date-time":"2026-09-01T00:00:00Z","timestamp":1788220800000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/www.elsevier.com\/legal\/tdmrep-license"},{"start":{"date-parts":[[2026,9,1]],"date-time":"2026-09-01T00:00:00Z","timestamp":1788220800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-017"},{"start":{"date-parts":[[2026,9,1]],"date-time":"2026-09-01T00:00:00Z","timestamp":1788220800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"},{"start":{"date-parts":[[2026,9,1]],"date-time":"2026-09-01T00:00:00Z","timestamp":1788220800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-012"},{"start":{"date-parts":[[2026,9,1]],"date-time":"2026-09-01T00:00:00Z","timestamp":1788220800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,9,1]],"date-time":"2026-09-01T00:00:00Z","timestamp":1788220800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-004"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62403360"],"award-info":[{"award-number":["62403360"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["92367101"],"award-info":[{"award-number":["92367101"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["elsevier.com","sciencedirect.com"],"crossmark-restriction":true},"short-container-title":["Neurocomputing"],"published-print":{"date-parts":[[2026,9]]},"DOI":"10.1016\/j.neucom.2026.133869","type":"journal-article","created":{"date-parts":[[2026,5,6]],"date-time":"2026-05-06T16:24:19Z","timestamp":1778084659000},"page":"133869","update-policy":"https:\/\/doi.org\/10.1016\/elsevier_cm_policy","source":"Crossref","is-referenced-by-count":0,"special_numbering":"C","title":["Understanding non-convex optimization in differentiable models via Fenchel\u2013Young loss: Theory and applications to deep learning"],"prefix":"10.1016","volume":"693","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-5832-1884","authenticated-orcid":false,"given":"Binchuan","family":"Qi","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4030-4968","authenticated-orcid":false,"given":"Wei","family":"Gong","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Li","family":"Li","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"78","reference":[{"key":"10.1016\/j.neucom.2026.133869_bib0005","series-title":"Proc. Adv. Neural Inf. Process. Syst","first-page":"687","article-title":"SPIDER: near-optimal non-convex optimization via stochastic path-integrated differential estimator","author":"Fang","year":"2018"},{"key":"10.1016\/j.neucom.2026.133869_bib0010","series-title":"Proc. Mach. Learn. Res volume 40 of JMLR Workshop and Conference Proceedings","first-page":"797","article-title":"Escaping from saddle points - online stochastic gradient for tensor decomposition","author":"Rong","year":"2015"},{"issue":"1","key":"10.1016\/j.neucom.2026.133869_bib0015","doi-asserted-by":"crossref","first-page":"455","DOI":"10.1093\/imanum\/drz055","article-title":"Strong error analysis for stochastic gradient descent optimization algorithms","volume":"41","author":"Jentzen","year":"2018","journal-title":"IMA J. Numer. Anal."},{"key":"10.1016\/j.neucom.2026.133869_bib0020","first-page":"2","article-title":"How to advance in structural convex optimization","volume":"78","author":"Nesterov","year":"2008","journal-title":"OPTIMA: Mathematical Programming Society Newsletter"},{"key":"10.1016\/j.neucom.2026.133869_bib0025","series-title":"Proc. IEEE Conf. Comput. Vis. Pattern Recognit","first-page":"770","article-title":"Deep residual learning for image recognition","author":"Kaiming","year":"2016"},{"key":"10.1016\/j.neucom.2026.133869_bib0030","series-title":"Proc. Int. Conf. Learn. Represent","article-title":"Gradient descent provably optimizes over-parameterized neural networks","author":"Simon","year":"2019"},{"key":"10.1016\/j.neucom.2026.133869_bib0035","series-title":"Proc. Adv. Neural Inf. Process. Syst","first-page":"37510","article-title":"Annihilation of spurious minima in two-layer RELU networks","volume":"vol. 35","author":"Arjevani","year":"2022"},{"key":"10.1016\/j.neucom.2026.133869_bib0040","series-title":"Bayesian Learning for Neural Networks","volume":"vol. 118","author":"Neal","year":"2012"},{"key":"10.1016\/j.neucom.2026.133869_bib0045","series-title":"Neural tangent kernel: convergence and generalization in neural networks (invited paper)","first-page":"6","author":"Jacot","year":"2021"},{"key":"10.1016\/j.neucom.2026.133869_bib0050","doi-asserted-by":"crossref","first-page":"6029","DOI":"10.1109\/TIP.2024.3478854","article-title":"Ntk-guided few-shot class incremental learning","volume":"33","author":"Liu","year":"2024","journal-title":"IEEE Trans. Image Process."},{"issue":"3","key":"10.1016\/j.neucom.2026.133869_bib0055","doi-asserted-by":"crossref","first-page":"1820","DOI":"10.1016\/j.spa.2019.06.003","article-title":"Mean field analysis of neural networks: a central limit theorem","volume":"130","author":"Sirignano","year":"2018","journal-title":"Stochastic Processes Appl."},{"key":"10.1016\/j.neucom.2026.133869_bib0060","series-title":"Proc. Mach. Learn. Res","first-page":"868","article-title":"Analyzing finite neural networks: can we trust neural tangent kernel theory?","author":"Seleznova","year":"2022"},{"key":"10.1016\/j.neucom.2026.133869_bib0065","series-title":"Proc. Int. Conf. Mach. Learn","first-page":"19522","article-title":"Neural tangent kernel beyond the infinite-width limit: effects of depth and initialization","author":"Seleznova","year":"2022"},{"key":"10.1016\/j.neucom.2026.133869_bib0070","article-title":"Empirical limitations of the NTK for understanding scaling laws in deep learning","volume":"2023","author":"Vyas","year":"2023","journal-title":"Trans. Mach. Learn. Res."},{"issue":"126227","key":"10.1016\/j.neucom.2026.133869_bib0075","article-title":"Do we really need a new theory to understand over-parameterization?","volume":"543","author":"Oneto","year":"2023","journal-title":"Neurocomputing"},{"key":"10.1016\/j.neucom.2026.133869_bib0080","series-title":"Proc. Adv. Neural Inf. Process. Syst","first-page":"2933","article-title":"Identifying and attacking the saddle point problem in high-dimensional non-convex optimization","author":"Dauphin","year":"2014"},{"key":"10.1016\/j.neucom.2026.133869_bib0085","series-title":"Proc. Int. Conf. Mach. Learn","first-page":"1019","article-title":"Sharp minima can generalize for deep nets","author":"Dinh","year":"2017"},{"issue":"35","key":"10.1016\/j.neucom.2026.133869_bib0090","first-page":"1","article-title":"Learning with fenchel-young losses","volume":"21","author":"Blondel","year":"2020","journal-title":"J. Mach. Learn. Res."},{"key":"10.1016\/j.neucom.2026.133869_bib0095","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1162\/neco.1997.9.1.1","article-title":"Flat minima","volume":"9","author":"Hochreiter","year":"1997","journal-title":"Neural Comput."},{"key":"10.1016\/j.neucom.2026.133869_bib0100","author":"Lewkowycz"},{"key":"10.1016\/j.neucom.2026.133869_bib0105","series-title":"Proc. Int. Conf. Learn. Represent","article-title":"Sharpness-aware minimization for efficiently improving generalization","author":"Foret","year":"2021"},{"key":"10.1016\/j.neucom.2026.133869_bib0110","series-title":"Proc. Adv. Neural Inf. Process. Syst","article-title":"Learning energy networks with generalized fenchel-young losses","author":"Blondel","year":"2022"},{"key":"10.1016\/j.neucom.2026.133869_bib0115","author":"Sklaviadis"},{"key":"10.1016\/j.neucom.2026.133869_bib0120","doi-asserted-by":"crossref","first-page":"18","DOI":"10.1109\/MSP.2020.3003845","article-title":"Understanding notions of stationarity in nonsmooth optimization: a guided tour of various constructions of subdifferential for nonsmooth functions","volume":"37","author":"Jiajin","year":"2020","journal-title":"IEEE Signal Process. Mag"},{"key":"10.1016\/j.neucom.2026.133869_bib0125","series-title":"Convex Analysis and Nonlinear Optimization: Theory and Examples","author":"Todd","year":"2003"},{"key":"10.1016\/j.neucom.2026.133869_bib0130","series-title":"Correction To: Convex Analysis and Monotone Operator Theory in Hilbert Spaces","author":"Bauschke","year":"2017"},{"key":"10.1016\/j.neucom.2026.133869_bib0135","series-title":"Proc. Int. Conf. Mach. Learn, Volume 70 of Proceedings of Machine Learning Research","first-page":"342","article-title":"The shattered gradients problem: if resnets are the answer, then what is the question?","author":"Balduzzi","year":"2017"},{"key":"10.1016\/j.neucom.2026.133869_bib0140","series-title":"Proc. Adv. Neural Inf. Process. Syst","first-page":"1106","article-title":"Imagenet classification with deep convolutional neural networks","author":"Krizhevsky","year":"2012"},{"issue":"11","key":"10.1016\/j.neucom.2026.133869_bib0145","doi-asserted-by":"crossref","first-page":"2278","DOI":"10.1109\/5.726791","article-title":"Gradient-based learning applied to document recognition","volume":"86","author":"LeCun","year":"1998","journal-title":"Proc. IEEE"},{"issue":"4","key":"10.1016\/j.neucom.2026.133869_bib0150","article-title":"Learning multiple layers of features from tiny images","volume":"1","author":"Krizhevsky","year":"2009","journal-title":"Handbook of Systemic Autoimmune Diseases"},{"key":"10.1016\/j.neucom.2026.133869_bib0155","series-title":"Proc. Adv. Neural Inf. Process. Syst","first-page":"1","article-title":"Going deeper with convolutions","author":"Szegedy","year":"2015"},{"key":"10.1016\/j.neucom.2026.133869_bib0160","series-title":"Proc. Eur. Conf. Comput. Vis","first-page":"630","article-title":"Identity mappings in deep residual networks","author":"Kaiming","year":"2016"},{"key":"10.1016\/j.neucom.2026.133869_bib0165","series-title":"Proc. IEEE Conf. Comput. Vis. Pattern Recognit","first-page":"4700","article-title":"Densely connected convolutional networks","author":"Huang","year":"2017"},{"key":"10.1016\/j.neucom.2026.133869_bib0170","author":"Howard"},{"key":"10.1016\/j.neucom.2026.133869_bib0175","series-title":"Proc. Eur. Conf. Comput. Vis","first-page":"116","article-title":"Shufflenet v2: practical guidelines for efficient CNN architecture design","author":"Ningning","year":"2018"},{"issue":"8","key":"10.1016\/j.neucom.2026.133869_bib0180","doi-asserted-by":"crossref","first-page":"2011","DOI":"10.1109\/TPAMI.2019.2913372","article-title":"Squeeze-and-excitation networks","volume":"42","author":"Jie","year":"2020","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"10.1016\/j.neucom.2026.133869_bib0185","author":"Dosovitskiy"},{"key":"10.1016\/j.neucom.2026.133869_bib0190","series-title":"Proc. IEEE Conf. Comput. Vis. Pattern Recognit","first-page":"14901","article-title":"Transformers without normalization","author":"Zhu","year":"2025"},{"key":"10.1016\/j.neucom.2026.133869_bib0195","series-title":"Proc. IEEE Int. Conf. Comput. Vis","first-page":"32","article-title":"Going deeper with image transformers","author":"Touvron","year":"2021"},{"key":"10.1016\/j.neucom.2026.133869_bib0200","author":"Yoshioka"},{"key":"10.1016\/j.neucom.2026.133869_bib0205","series-title":"Of Applied Optimization","doi-asserted-by":"crossref","DOI":"10.1007\/978-1-4419-8853-9","article-title":"Introductory lectures on convex optimization - a basic course","volume":"vol. 87","author":"Nesterov","year":"2004"},{"issue":"1","key":"10.1016\/j.neucom.2026.133869_bib0210","doi-asserted-by":"crossref","first-page":"71","DOI":"10.1007\/s10107-019-01406-y","article-title":"Lower bounds for finding stationary points i","volume":"184","author":"Carmon","year":"2020","journal-title":"Math. Program."},{"issue":"1","key":"10.1016\/j.neucom.2026.133869_bib0215","doi-asserted-by":"crossref","first-page":"165","DOI":"10.1007\/s10107-022-01822-7","article-title":"Lower bounds for non-convex stochastic optimization","volume":"199","author":"Arjevani","year":"2023","journal-title":"Math. Program."},{"issue":"87\u201389","key":"10.1016\/j.neucom.2026.133869_bib0220","article-title":"Une propri\u00e9t\u00e9 topologique des sous-ensembles analytiques r\u00e9els","volume":"117","author":"Lojasiewicz","year":"1963","journal-title":"Les \u00e9quations aux d\u00e9riv\u00e9es partielles"},{"issue":"3","key":"10.1016\/j.neucom.2026.133869_bib0225","doi-asserted-by":"crossref","first-page":"769","DOI":"10.5802\/aif.1638","article-title":"On gradients of functions definable in o-minimal structures","volume":"48","author":"Kurdyka","year":"1998","journal-title":"Ann. Inst. Fourier"},{"issue":"2","key":"10.1016\/j.neucom.2026.133869_bib0230","doi-asserted-by":"crossref","first-page":"556","DOI":"10.1137\/060670080","article-title":"Clarke subgradients of stratifiable functions","volume":"18","author":"Bolte","year":"2007","journal-title":"SIAM J. Optim."},{"issue":"4","key":"10.1016\/j.neucom.2026.133869_bib0235","doi-asserted-by":"crossref","first-page":"2765","DOI":"10.1287\/moor.2021.1227","article-title":"The exact modulus of the generalized concave kurdyka-\u0142ojasiewicz property","volume":"47","author":"Wang","year":"2022","journal-title":"Math. Oper. Res."},{"key":"10.1016\/j.neucom.2026.133869_bib0240","series-title":"Elements of Information Theory (2. Ed.)","author":"Cover","year":"2006"},{"key":"10.1016\/j.neucom.2026.133869_bib0245","first-page":"299","article-title":"On information-type measure of difference of probability distributions and indirect observations","volume":"2","author":"Csisz\u00e1r","year":"1967","journal-title":"Studia Sci. Math. Hungar."},{"key":"10.1016\/j.neucom.2026.133869_bib0250","doi-asserted-by":"crossref","first-page":"126","DOI":"10.1109\/TIT.1967.1053968","article-title":"A lower bound for discrimination information in terms of variation (corresp.)","volume":"13","author":"Kullback","year":"1967","journal-title":"IEEE Trans. Inf. Theory"},{"issue":"3","key":"10.1016\/j.neucom.2026.133869_bib0255","doi-asserted-by":"crossref","first-page":"1007","DOI":"10.1109\/TIT.2005.864431","article-title":"Context tree estimation for not necessarily finite memory processes, via bic and mdl","volume":"52","author":"Csiszar","year":"2006","journal-title":"IEEE Trans. Inf. Theory"},{"key":"10.1016\/j.neucom.2026.133869_bib0260","series-title":"Foundations of Data Science","author":"Blum","year":"2020"}],"container-title":["Neurocomputing"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/api.elsevier.com\/content\/article\/PII:S092523122601266X?httpAccept=text\/xml","content-type":"text\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/api.elsevier.com\/content\/article\/PII:S092523122601266X?httpAccept=text\/plain","content-type":"text\/plain","content-version":"vor","intended-application":"text-mining"}],"deposited":{"date-parts":[[2026,7,3]],"date-time":"2026-07-03T02:35:32Z","timestamp":1783046132000},"score":1,"resource":{"primary":{"URL":"https:\/\/linkinghub.elsevier.com\/retrieve\/pii\/S092523122601266X"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9]]},"references-count":52,"alternative-id":["S092523122601266X"],"URL":"https:\/\/doi.org\/10.1016\/j.neucom.2026.133869","relation":{},"ISSN":["0925-2312"],"issn-type":[{"value":"0925-2312","type":"print"}],"subject":[],"published":{"date-parts":[[2026,9]]},"assertion":[{"value":"Elsevier","name":"publisher","label":"This article is maintained by"},{"value":"Understanding non-convex optimization in differentiable models via Fenchel\u2013Young loss: Theory and applications to deep learning","name":"articletitle","label":"Article Title"},{"value":"Neurocomputing","name":"journaltitle","label":"Journal Title"},{"value":"https:\/\/doi.org\/10.1016\/j.neucom.2026.133869","name":"articlelink","label":"CrossRef DOI link to publisher maintained version"},{"value":"article","name":"content_type","label":"Content Type"},{"value":"\u00a9 2026 Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.","name":"copyright","label":"Copyright"}],"article-number":"133869"}}