{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,11]],"date-time":"2026-05-11T23:08:30Z","timestamp":1778540910745,"version":"3.51.4"},"reference-count":46,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"6","license":[{"start":{"date-parts":[[2021,6,1]],"date-time":"2021-06-01T00:00:00Z","timestamp":1622505600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2021,6,1]],"date-time":"2021-06-01T00:00:00Z","timestamp":1622505600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2021,6,1]],"date-time":"2021-06-01T00:00:00Z","timestamp":1622505600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/100000181","name":"Air Force Office of Scientific Research","doi-asserted-by":"publisher","award":["FA9550-20-1-0122"],"award-info":[{"award-number":["FA9550-20-1-0122"]}],"id":[{"id":"10.13039\/100000181","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100000001","name":"NSF","doi-asserted-by":"publisher","award":["1932196"],"award-info":[{"award-number":["1932196"]}],"id":[{"id":"10.13039\/100000001","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100004316","name":"Rensselaer-IBM AI Research Collaboration","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100004316","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100004316","name":"IBM AI Horizons Network","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100004316","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Neural Netw. Learning Syst."],"published-print":{"date-parts":[[2021,6]]},"DOI":"10.1109\/tnnls.2020.3007399","type":"journal-article","created":{"date-parts":[[2020,7,29]],"date-time":"2020-07-29T20:42:52Z","timestamp":1596055372000},"page":"2622-2635","source":"Crossref","is-referenced-by-count":11,"title":["Improved Linear Convergence of Training CNNs With Generalizability Guarantees: A One-Hidden-Layer Case"],"prefix":"10.1109","volume":"32","author":[{"given":"Shuai","family":"Zhang","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0928-9691","authenticated-orcid":false,"given":"Meng","family":"Wang","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2620-4859","authenticated-orcid":false,"given":"Jinjun","family":"Xiong","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2817-6991","authenticated-orcid":false,"given":"Sijia","family":"Liu","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Pin-Yu","family":"Chen","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1109\/TSP.2019.2904921"},{"key":"ref38","article-title":"Introduction to the non-asymptotic analysis of random matrices","author":"vershynin","year":"2010","journal-title":"arXiv 1011 3027"},{"key":"ref33","doi-asserted-by":"crossref","first-page":"484","DOI":"10.1038\/nature16961","article-title":"Mastering the game of Go with deep neural networks and tree search","volume":"529","author":"silver","year":"2016","journal-title":"Nature"},{"key":"ref32","first-page":"1135","article-title":"Distribution-specific hardness of learning neural networks","volume":"19","author":"shamir","year":"2018","journal-title":"J Mach Learn Res"},{"key":"ref31","first-page":"4430","article-title":"Spurious local minima are common in two-layer ReLU neural networks","author":"safran","year":"2018","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref30","first-page":"533","article-title":"Learning representations by back-propagating errors","volume":"5","author":"rumelhart","year":"1988","journal-title":"Modeling Cognition"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1007\/s10208-011-9099-z"},{"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":"70","author":"tian","year":"2017","journal-title":"Proc 34th Int Conf Mach Learn"},{"key":"ref35","first-page":"1139","article-title":"On the importance of initialization and momentum in deep learning","author":"sutskever","year":"2013","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/TIT.2018.2854560"},{"key":"ref10","first-page":"1","article-title":"Learning one-hidden-layer neural networks with landscape design","author":"ge","year":"2018","journal-title":"Proc Int Conf Learn Represent"},{"key":"ref40","first-page":"8105","article-title":"Learning distributions generated by one-layer ReLU networks","author":"wu","year":"2019","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref11","first-page":"315","article-title":"Deep sparse rectifier neural networks","author":"glorot","year":"2011","journal-title":"Proc 14th Int Conf Artif Intell Statist"},{"key":"ref12","first-page":"1783","article-title":"Learning one convolutional layer with overlapping patches","author":"goel","year":"2018","journal-title":"Proc ICML"},{"key":"ref13","first-page":"217","article-title":"Permitted and forbidden sets in symmetric threshold-linear networks","author":"hahnloser","year":"2001","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref14","first-page":"1225","article-title":"Train faster, generalize better: Stability of stochastic gradient descent","author":"hardt","year":"2016","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref15","article-title":"On large-batch training for deep learning: Generalization gap and sharp minima","author":"keskar","year":"2016","journal-title":"arXiv 1609 04836"},{"key":"ref16","first-page":"1097","article-title":"Imagenet classification with deep convolutional neural networks","author":"krizhevsky","year":"2012","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref17","first-page":"507","article-title":"Tensor factorization via matrix factorization","author":"kuleshov","year":"2015","journal-title":"Proc Artif Intell Statist"},{"key":"ref18","first-page":"2908","article-title":"The multilinear structure of ReLU networks","author":"laurent","year":"2018","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/72.554195"},{"key":"ref28","author":"polyak","year":"1987","journal-title":"Introduction to Optimization"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1109\/TGRS.2016.2584107"},{"key":"ref27","first-page":"5947","article-title":"Exploring generalization in deep learning","author":"neyshabur","year":"2017","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref3","first-page":"605","article-title":"Globally optimal gradient descent for a ConvNet with Gaussian inputs","volume":"70","author":"brutzkus","year":"2017","journal-title":"Proc 34th Int Conf Mach Learn"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2009.5206848"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1037\/h0042519"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1145\/1390156.1390177"},{"key":"ref8","first-page":"1338","article-title":"Gradient descent learns one-hidden-layer CNN: Don’t be afraid of spurious local minima","author":"du","year":"2018","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref7","article-title":"When is a convolutional filter easy to learn?","author":"du","year":"2017","journal-title":"arXiv 1709 06129"},{"key":"ref2","first-page":"499","article-title":"Stability and generalization","volume":"2","author":"bousquet","year":"2002","journal-title":"J Mach Learn Res"},{"key":"ref9","article-title":"Guaranteed recovery of one-hidden-layer neural networks via cross entropy","author":"fu","year":"2018","journal-title":"arXiv 1802 06463"},{"key":"ref1","first-page":"6155","article-title":"Learning and generalization in overparameterized neural networks, going beyond two layers","author":"allen-zhu","year":"2019","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref46","first-page":"4140","article-title":"Recovery guarantees for one-hidden-layer neural networks","volume":"70","author":"zhong","year":"2017","journal-title":"Proc 34th Int Conf Mach Learn"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/5.726791"},{"key":"ref45","article-title":"Learning non-overlapping convolutional neural networks with multiple kernels","author":"zhong","year":"2017","journal-title":"arXiv 1711 03440"},{"key":"ref22","first-page":"597","article-title":"Convergence analysis of two-layer neural networks with ReLU activation","author":"li","year":"2017","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1038\/nature14539"},{"key":"ref42","first-page":"6594","article-title":"On the power and limitations of random features for understanding neural networks","author":"yehudai","year":"2019","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref24","first-page":"855","article-title":"On the computational efficiency of training neural networks","author":"livni","year":"2014","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref41","article-title":"Unified convergence analysis of stochastic momentum methods for convex and non-convex optimization","author":"yang","year":"2016","journal-title":"arXiv 1604 03257"},{"key":"ref23","first-page":"4355","article-title":"Adding one neuron can eliminate all bad local minima","author":"liang","year":"2018","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref44","first-page":"1524","article-title":"Learning one-hidden-layer ReLU networks via gradient descent","author":"zhang","year":"2019","journal-title":"Proc 22nd Int Conf Artif Intell Statist"},{"key":"ref26","volume":"87","author":"nesterov","year":"2013","journal-title":"Introductory Lectures on Convex Optimization A Basic Course"},{"key":"ref43","article-title":"Understanding deep learning requires rethinking generalization","author":"zhang","year":"2016","journal-title":"arXiv 1611 03530"},{"key":"ref25","first-page":"1","article-title":"Rectifier nonlinearities improve neural network acoustic models","volume":"30","author":"maas","year":"2013","journal-title":"Proc 30th Int Conf Mach Learn"}],"container-title":["IEEE Transactions on Neural Networks and Learning Systems"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/5962385\/9445711\/09152147.pdf?arnumber=9152147","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,5,10]],"date-time":"2022-05-10T14:53:15Z","timestamp":1652194395000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/9152147\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021,6]]},"references-count":46,"journal-issue":{"issue":"6"},"URL":"https:\/\/doi.org\/10.1109\/tnnls.2020.3007399","relation":{},"ISSN":["2162-237X","2162-2388"],"issn-type":[{"value":"2162-237X","type":"print"},{"value":"2162-2388","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,6]]}}}