{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,10,28]],"date-time":"2025-10-28T15:11:23Z","timestamp":1761664283128,"version":"3.37.3"},"reference-count":41,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","license":[{"start":{"date-parts":[[2021,1,1]],"date-time":"2021-01-01T00:00:00Z","timestamp":1609459200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2021,1,1]],"date-time":"2021-01-01T00:00:00Z","timestamp":1609459200000},"content-version":"am","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2021,1,1]],"date-time":"2021-01-01T00:00:00Z","timestamp":1609459200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2021,1,1]],"date-time":"2021-01-01T00:00:00Z","timestamp":1609459200000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["61925303","62088101","U20B2073","61720106011"],"award-info":[{"award-number":["61925303","62088101","U20B2073","61720106011"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100013290","name":"National Key R&D Program of China","doi-asserted-by":"publisher","award":["2018YFB1700100"],"award-info":[{"award-number":["2018YFB1700100"]}],"id":[{"id":"10.13039\/501100013290","id-type":"DOI","asserted-by":"publisher"}]},{"name":"NSF","award":["1901134"],"award-info":[{"award-number":["1901134"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Signal Process."],"published-print":{"date-parts":[[2021]]},"DOI":"10.1109\/tsp.2021.3094911","type":"journal-article","created":{"date-parts":[[2021,7,7]],"date-time":"2021-07-07T19:36:48Z","timestamp":1625686608000},"page":"4416-4427","source":"Crossref","is-referenced-by-count":11,"title":["Learning Two-Layer ReLU Networks Is Nearly as Easy as Learning Linear Classifiers on Separable Data"],"prefix":"10.1109","volume":"69","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-7789-6552","authenticated-orcid":false,"given":"Qiuling","family":"Yang","sequence":"first","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0003-1280-7592","authenticated-orcid":false,"given":"Alireza","family":"Sadeghi","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7266-2412","authenticated-orcid":false,"given":"Gang","family":"Wang","sequence":"additional","affiliation":[]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9898-3129","authenticated-orcid":false,"given":"Jian","family":"Sun","sequence":"additional","affiliation":[]}],"member":"263","reference":[{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1109\/CISS48834.2020.1570627111"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1109\/TSP.2019.2926023"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1109\/TNNLS.2013.2246578"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1109\/TIT.2017.2756858"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1109\/TSP.2019.2904921"},{"key":"ref30","first-page":"25","article-title":"Max-margin Markov networks","author":"taskar","year":"2004","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref37","first-page":"1","article-title":"Understanding deep learning requires rethinking generalization","author":"zhang","year":"2017","journal-title":"Proc Int Conf Learn Representation"},{"key":"ref36","first-page":"1","article-title":"Efficiently testing local optimality and escaping saddles for ReLU networks","author":"yun","year":"2019","journal-title":"Proc Int Learn Representation"},{"article-title":"Small nonlinearities in activation functions create bad local minima in neural networks","year":"2018","author":"yun","key":"ref35"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1109\/TSG.2019.2951769"},{"article-title":"Approximation schemes for ReLU regression","year":"2020","author":"diakonikolas","key":"ref10"},{"article-title":"From symmetry to geometry: Tractable nonconvex problems","year":"2020","author":"zhang","key":"ref40"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1109\/TSP.2020.2993153"},{"key":"ref12","first-page":"297","article-title":"Size-independent sample complexity of neural networks","author":"golowich","year":"2018","journal-title":"Proc Conf Learn Theory"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-73074-5_5"},{"key":"ref14","first-page":"586","article-title":"Deep learning without poor local minima","author":"kawaguchi","year":"2016","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref15","first-page":"2908","article-title":"The multilinear structure of ReLU networks","author":"laurent","year":"2018","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref16","first-page":"2902","article-title":"Deep linear networks with arbitrary loss: All local minima are global","author":"laurent","year":"2018","journal-title":"Proc Int Conf Mach Learn"},{"key":"ref17","doi-asserted-by":"crossref","first-page":"436","DOI":"10.1038\/nature14539","article-title":"Deep learning","volume":"521","author":"lecun","year":"2015","journal-title":"Nature"},{"article-title":"Over-parameterized deep neural networks have no strict local minima for any continuous activations","year":"2018","author":"li","key":"ref18"},{"key":"ref19","first-page":"8157","article-title":"Learning overparameterized neural networks via stochastic gradient descent on structured data","author":"li","year":"2018","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1109\/TIT.2018.2854560"},{"key":"ref4","first-page":"6240","article-title":"Spectrally-normalized margin bounds for neural networks","author":"bartlett","year":"2017","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref27","first-page":"2007","article-title":"Learning ReLUs via gradient descent","author":"soltanolkotabi","year":"2017","journal-title":"Proc Adv Neural Inf Process Syst"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP40776.2020.9053864"},{"key":"ref6","first-page":"1","article-title":"SGD learns over-parameterized networks that provably generalize on linearly separable data","author":"brutzkus","year":"2018","journal-title":"Proc Int Conf Learn Rep"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1109\/TIT.2016.2598574"},{"key":"ref5","first-page":"1","article-title":"Training a 3-node neural network is NP-complete","author":"blum","year":"1988","journal-title":"Proc Adv Neural Inf Process Syst"},{"article-title":"How much over-parameterization is sufficient to learn deep ReLU networks","year":"2020","author":"chen","key":"ref8"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1109\/MSP.2020.2984780"},{"article-title":"Fine-grained analysis of optimization and generalization for overparameterized two-layer neural networks","year":"2019","author":"arora","key":"ref2"},{"article-title":"UCI machine learning repository","year":"2017","author":"dheeru","key":"ref9"},{"article-title":"A convergence theory for deep learning via over-parameterization","year":"2018","author":"allen-zhu","key":"ref1"},{"article-title":"Algorithmic regularization in over-parameterized matrix sensing and neural networks with quadratic activations","year":"2017","author":"li","key":"ref20"},{"key":"ref22","first-page":"3730","author":"nguyen","year":"2018","journal-title":"Proc Int Conf Mach Learn"},{"article-title":"Revisiting landscape analysis in deep neural networks: Eliminating decreasing paths to infinity","year":"2019","author":"liang","key":"ref21"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1109\/JSAIT.2020.2991332"},{"key":"ref41","first-page":"1","article-title":"Critical points of linear neural networks: Analytical forms and landscape properties","author":"zhou","year":"2018","journal-title":"Proc Int Conf Learn Repr"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/IEEECONF44664.2019.9048845"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1017\/CBO9781107298019"},{"key":"ref25","first-page":"4433","article-title":"Spurious local minima are common in two-layer ReLU neural networks","author":"safran","year":"2018","journal-title":"Proc Int Conf Mach Learn"}],"container-title":["IEEE Transactions on Signal Processing"],"original-title":[],"link":[{"URL":"https:\/\/ieeexplore.ieee.org\/ielam\/78\/9307529\/9477126-aam.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/78\/9307529\/09477126.pdf?arnumber=9477126","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,5,10]],"date-time":"2022-05-10T14:50:36Z","timestamp":1652194236000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/9477126\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021]]},"references-count":41,"URL":"https:\/\/doi.org\/10.1109\/tsp.2021.3094911","relation":{},"ISSN":["1053-587X","1941-0476"],"issn-type":[{"type":"print","value":"1053-587X"},{"type":"electronic","value":"1941-0476"}],"subject":[],"published":{"date-parts":[[2021]]}}}