{"id":"https://openalex.org/W3134450292","doi":"https://doi.org/10.1109/tsp.2022.3156702","title":"Self-Regularity of Non-Negative Output Weights for Overparameterized Two-Layer Neural Networks","display_name":"Self-Regularity of Non-Negative Output Weights for Overparameterized Two-Layer Neural Networks","publication_year":2022,"publication_date":"2022-01-01","ids":{"openalex":"https://openalex.org/W3134450292","doi":"https://doi.org/10.1109/tsp.2022.3156702","mag":"3134450292"},"language":"en","primary_location":{"id":"doi:10.1109/tsp.2022.3156702","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tsp.2022.3156702","pdf_url":null,"source":{"id":"https://openalex.org/S168680287","display_name":"IEEE Transactions on Signal Processing","issn_l":"1053-587X","issn":["1053-587X","1941-0476"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Signal Processing","raw_type":"journal-article"},"type":"article","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2103.01887","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5031648070","display_name":"David Gamarnik","orcid":null},"institutions":[{"id":"https://openalex.org/I4210110987","display_name":"IIT@MIT","ror":"https://ror.org/01wp8zh54","country_code":"US","type":"facility","lineage":["https://openalex.org/I30771326","https://openalex.org/I4210110987"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"David Gamarnik","raw_affiliation_strings":["MIT, Cambridge, MA, USA"],"raw_orcid":"https://orcid.org/0000-0001-8898-8778","affiliations":[{"raw_affiliation_string":"MIT, Cambridge, MA, USA","institution_ids":["https://openalex.org/I4210110987"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5029207194","display_name":"Eren C. K\u0131z\u0131lda\u011f","orcid":"https://orcid.org/0000-0003-0411-7161"},"institutions":[{"id":"https://openalex.org/I4210110987","display_name":"IIT@MIT","ror":"https://ror.org/01wp8zh54","country_code":"US","type":"facility","lineage":["https://openalex.org/I30771326","https://openalex.org/I4210110987"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Eren C. Kzldag","raw_affiliation_strings":["MIT, Cambridge, MA, USA"],"raw_orcid":"https://orcid.org/0000-0003-0411-7161","affiliations":[{"raw_affiliation_string":"MIT, Cambridge, MA, USA","institution_ids":["https://openalex.org/I4210110987"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5061028424","display_name":"Ilias Zadik","orcid":"https://orcid.org/0009-0008-4085-324X"},"institutions":[{"id":"https://openalex.org/I102475099","display_name":"Sup\u00e9lec","ror":"https://ror.org/00n7gwn90","country_code":"FR","type":"education","lineage":["https://openalex.org/I102475099"]},{"id":"https://openalex.org/I15196421","display_name":"University of Applied Sciences and Arts of Southern Switzerland","ror":"https://ror.org/05ep8g269","country_code":"CH","type":"education","lineage":["https://openalex.org/I15196421"]},{"id":"https://openalex.org/I4210102264","display_name":"Shandong University of Political Science and Law","ror":"https://ror.org/01b2j5886","country_code":"CN","type":"education","lineage":["https://openalex.org/I4210102264"]},{"id":"https://openalex.org/I57206974","display_name":"New York University","ror":"https://ror.org/0190ak572","country_code":"US","type":"education","lineage":["https://openalex.org/I57206974"]}],"countries":["CH","CN","FR","US"],"is_corresponding":false,"raw_author_name":"Ilias Zadik","raw_affiliation_strings":["New York University, New York, NY, USA","New York University<sup>**</sup>"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"New York University, New York, NY, USA","institution_ids":["https://openalex.org/I57206974"]},{"raw_affiliation_string":"New York University<sup>**</sup>","institution_ids":["https://openalex.org/I102475099","https://openalex.org/I15196421","https://openalex.org/I4210102264"]}]}],"institutions":[],"countries_distinct_count":4,"institutions_distinct_count":5,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.1161,"has_fulltext":false,"cited_by_count":3,"citation_normalized_percentile":{"value":0.48726317,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":89,"max":97},"biblio":{"volume":"70","issue":null,"first_page":"1310","last_page":"1319"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9991000294685364,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9991000294685364,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T12072","display_name":"Machine Learning and Algorithms","score":0.9990000128746033,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T12676","display_name":"Machine Learning and ELM","score":0.9979000091552734,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.5834919214248657},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.562807559967041},{"id":"https://openalex.org/keywords/leverage","display_name":"Leverage (statistics)","score":0.540816605091095},{"id":"https://openalex.org/keywords/vc-dimension","display_name":"VC dimension","score":0.5338571667671204},{"id":"https://openalex.org/keywords/norm","display_name":"Norm (philosophy)","score":0.5288558602333069},{"id":"https://openalex.org/keywords/sigmoid-function","display_name":"Sigmoid function","score":0.4925768971443176},{"id":"https://openalex.org/keywords/generalization","display_name":"Generalization","score":0.4836881756782532},{"id":"https://openalex.org/keywords/discrete-mathematics","display_name":"Discrete mathematics","score":0.4787939488887787},{"id":"https://openalex.org/keywords/dimension","display_name":"Dimension (graph theory)","score":0.4428607225418091},{"id":"https://openalex.org/keywords/generalization-error","display_name":"Generalization error","score":0.44181859493255615},{"id":"https://openalex.org/keywords/time-complexity","display_name":"Time complexity","score":0.4358506500720978},{"id":"https://openalex.org/keywords/binary-number","display_name":"Binary number","score":0.42486006021499634},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.3986494541168213},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.38184720277786255},{"id":"https://openalex.org/keywords/combinatorics","display_name":"Combinatorics","score":0.3271304666996002},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.20554789900779724},{"id":"https://openalex.org/keywords/arithmetic","display_name":"Arithmetic","score":0.1980694830417633},{"id":"https://openalex.org/keywords/statistics","display_name":"Statistics","score":0.09960263967514038}],"concepts":[{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.5834919214248657},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.562807559967041},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.540816605091095},{"id":"https://openalex.org/C119322782","wikidata":"https://www.wikidata.org/wiki/Q2662236","display_name":"VC dimension","level":2,"score":0.5338571667671204},{"id":"https://openalex.org/C191795146","wikidata":"https://www.wikidata.org/wiki/Q3878446","display_name":"Norm (philosophy)","level":2,"score":0.5288558602333069},{"id":"https://openalex.org/C81388566","wikidata":"https://www.wikidata.org/wiki/Q526668","display_name":"Sigmoid function","level":3,"score":0.4925768971443176},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.4836881756782532},{"id":"https://openalex.org/C118615104","wikidata":"https://www.wikidata.org/wiki/Q121416","display_name":"Discrete mathematics","level":1,"score":0.4787939488887787},{"id":"https://openalex.org/C33676613","wikidata":"https://www.wikidata.org/wiki/Q13415176","display_name":"Dimension (graph theory)","level":2,"score":0.4428607225418091},{"id":"https://openalex.org/C117765406","wikidata":"https://www.wikidata.org/wiki/Q5362437","display_name":"Generalization error","level":3,"score":0.44181859493255615},{"id":"https://openalex.org/C311688","wikidata":"https://www.wikidata.org/wiki/Q2393193","display_name":"Time complexity","level":2,"score":0.4358506500720978},{"id":"https://openalex.org/C48372109","wikidata":"https://www.wikidata.org/wiki/Q3913","display_name":"Binary number","level":2,"score":0.42486006021499634},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.3986494541168213},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.38184720277786255},{"id":"https://openalex.org/C114614502","wikidata":"https://www.wikidata.org/wiki/Q76592","display_name":"Combinatorics","level":1,"score":0.3271304666996002},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.20554789900779724},{"id":"https://openalex.org/C94375191","wikidata":"https://www.wikidata.org/wiki/Q11205","display_name":"Arithmetic","level":1,"score":0.1980694830417633},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.09960263967514038},{"id":"https://openalex.org/C199539241","wikidata":"https://www.wikidata.org/wiki/Q7748","display_name":"Law","level":1,"score":0.0},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.0},{"id":"https://openalex.org/C17744445","wikidata":"https://www.wikidata.org/wiki/Q36442","display_name":"Political science","level":0,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1109/tsp.2022.3156702","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tsp.2022.3156702","pdf_url":null,"source":{"id":"https://openalex.org/S168680287","display_name":"IEEE Transactions on Signal Processing","issn_l":"1053-587X","issn":["1053-587X","1941-0476"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Signal Processing","raw_type":"journal-article"},{"id":"pmh:oai:arXiv.org:2103.01887","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2103.01887","pdf_url":"https://arxiv.org/pdf/2103.01887","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2103.01887","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2103.01887","pdf_url":"https://arxiv.org/pdf/2103.01887","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":124,"referenced_works":["https://openalex.org/W143004564","https://openalex.org/W1720114023","https://openalex.org/W1811750039","https://openalex.org/W1902027874","https://openalex.org/W1993882792","https://openalex.org/W2017753243","https://openalex.org/W2084544490","https://openalex.org/W2099579348","https://openalex.org/W2116216716","https://openalex.org/W2117130368","https://openalex.org/W2129192653","https://openalex.org/W2131172946","https://openalex.org/W2135029798","https://openalex.org/W2156387975","https://openalex.org/W2163605009","https://openalex.org/W2194775991","https://openalex.org/W2507533531","https://openalex.org/W2519994964","https://openalex.org/W2566079294","https://openalex.org/W2576544140","https://openalex.org/W2591714514","https://openalex.org/W2596625124","https://openalex.org/W2604117713","https://openalex.org/W2709553318","https://openalex.org/W2732724430","https://openalex.org/W2736970300","https://openalex.org/W2741952635","https://openalex.org/W2765861987","https://openalex.org/W2766371994","https://openalex.org/W2766447205","https://openalex.org/W2767382501","https://openalex.org/W2777138330","https://openalex.org/W2777256551","https://openalex.org/W2785631679","https://openalex.org/W2785749378","https://openalex.org/W2787248994","https://openalex.org/W2798418464","https://openalex.org/W2800415562","https://openalex.org/W2808958252","https://openalex.org/W2894604724","https://openalex.org/W2899790086","https://openalex.org/W2911867426","https://openalex.org/W2946510242","https://openalex.org/W2946840143","https://openalex.org/W2949804919","https://openalex.org/W2950220847","https://openalex.org/W2952318479","https://openalex.org/W2962698540","https://openalex.org/W2962857907","https://openalex.org/W2962930448","https://openalex.org/W2963038205","https://openalex.org/W2963236897","https://openalex.org/W2963239103","https://openalex.org/W2963285844","https://openalex.org/W2963383839","https://openalex.org/W2963508863","https://openalex.org/W2963518130","https://openalex.org/W2963664410","https://openalex.org/W2963672337","https://openalex.org/W2963739978","https://openalex.org/W2963794891","https://openalex.org/W2963826371","https://openalex.org/W2964310974","https://openalex.org/W2965497096","https://openalex.org/W2965677819","https://openalex.org/W2970332347","https://openalex.org/W2971043187","https://openalex.org/W2973737673","https://openalex.org/W2974929660","https://openalex.org/W2978073653","https://openalex.org/W3013992185","https://openalex.org/W3036841168","https://openalex.org/W3081446941","https://openalex.org/W3093317808","https://openalex.org/W3113714439","https://openalex.org/W3119586787","https://openalex.org/W3137695714","https://openalex.org/W3143596294","https://openalex.org/W3198742493","https://openalex.org/W4245577611","https://openalex.org/W4250954493","https://openalex.org/W4252305909","https://openalex.org/W4289436733","https://openalex.org/W4293448676","https://openalex.org/W4297738147","https://openalex.org/W6637414576","https://openalex.org/W6638214083","https://openalex.org/W6680012447","https://openalex.org/W6681017033","https://openalex.org/W6682889407","https://openalex.org/W6684191040","https://openalex.org/W6733941281","https://openalex.org/W6734855675","https://openalex.org/W6735544424","https://openalex.org/W6739659843","https://openalex.org/W6740483536","https://openalex.org/W6741472820","https://openalex.org/W6741653254","https://openalex.org/W6744989570","https://openalex.org/W6745448519","https://openalex.org/W6745751660","https://openalex.org/W6746713790","https://openalex.org/W6748256991","https://openalex.org/W6748600614","https://openalex.org/W6748748765","https://openalex.org/W6750531877","https://openalex.org/W6750606353","https://openalex.org/W6751002639","https://openalex.org/W6752544856","https://openalex.org/W6752591435","https://openalex.org/W6753918066","https://openalex.org/W6755150206","https://openalex.org/W6756001544","https://openalex.org/W6758902322","https://openalex.org/W6761496057","https://openalex.org/W6763368363","https://openalex.org/W6766400548","https://openalex.org/W6767324501","https://openalex.org/W6767695172","https://openalex.org/W6767838950","https://openalex.org/W6767902025","https://openalex.org/W6775446939","https://openalex.org/W6780065278","https://openalex.org/W6784331473"],"related_works":["https://openalex.org/W1494863055","https://openalex.org/W1993514889","https://openalex.org/W4295004871","https://openalex.org/W2964020133","https://openalex.org/W1714001321","https://openalex.org/W2949777805","https://openalex.org/W4225522981","https://openalex.org/W4321151660","https://openalex.org/W2524965963","https://openalex.org/W2009557879"],"abstract_inverted_index":{"We":[0,158],"consider":[1],"the":[2,34,39,48,51,64,85,120,135,145,149,176,180],"problem":[3],"of":[4,50,66,119,122,175,210],"finding":[5],"a":[6,23,42,71,92,102,111,172,198],"two-layer":[7],"neural":[8],"network":[9,90,181],"with":[10],"sigmoid,":[11],"rectified":[12],"linear":[13],"unit":[14],"(ReLU),":[15],"or":[16],"binary":[17],"step":[18],"activation":[19],"functions":[20],"that":[21,47,78,179],"\u201cfits\u201d":[22],"training":[24,35,44,94,136],"data":[25,99,146],"set":[26],"as":[27,29,31],"accurately":[28],"possible":[30],"quantified":[32],"by":[33],"error;":[36],"and":[37,138,201],"study":[38],"following":[40],"question:does":[41],"low":[43,199],"error":[45,95],"guarantee":[46],"norm":[49],"output":[52,68],"layer":[53],"(outer":[54],"norm)":[55],"itself":[56],"is":[57],"small?We":[58],"answer":[59],"affirmatively":[60],"this":[61],"question":[62],"for":[63,167,206],"case":[65],"non-negative":[67],"weights.":[69],"Using":[70],"simple":[72],"covering":[73],"number":[74,121],"argument,":[75],"we":[76,183],"establish":[77,164],"under":[79],"quite":[80,141],"mild":[81,142],"distributional":[82],"assumptions":[83,143],"on":[84,96,144],"input/label":[86],"pairs;":[87],"any":[88],"such":[89,168],"achieving":[91],"small":[93],"polynomially":[97],"many":[98],"necessarily":[100],"has":[101],"well-controlled":[103],"outer":[104],"norm.":[105],"Notably,":[106,187],"our":[107,161,188],"results":[108],"(a)":[109],"have":[110,155,192],"polynomial":[112],"(in$d$)":[113],"sample":[114,194],"complexity,":[115],"(b)":[116],"are":[117,132,202],"independent":[118,156],"hidden":[123],"units":[124],"(which":[125],"can":[126],"potentially":[127],"be":[128],"very":[129],"high),":[130],"(c)":[131],"oblivious":[133],"to":[134,163],"algorithm;":[137],"(d)":[139],"require":[140],"(in":[147],"particular":[148],"input":[150],"vector$X\\in":[151],"\\mathbb":[152],"{R}^{d}$need":[153],"not":[154],"coordinates).":[157],"then":[159],"leverage":[160],"bounds":[162,190],"generalization":[165,189],"guarantees":[166],"networks":[169],"throughfat-shattering":[170],"dimension,":[171],"scale-sensitive":[173],"measure":[174],"complexity":[177,195],"class":[178],"architectures":[182],"investigate":[184],"belong":[185],"to.":[186],"also":[191],"good":[193],"(polynomials":[196],"in$d$with":[197],"degree),":[200],"in":[203],"fact":[204],"near-linear":[205],"some":[207],"important":[208],"cases":[209],"interest.":[211]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2024,"cited_by_count":1},{"year":2021,"cited_by_count":1}],"updated_date":"2026-07-29T09:40:50.615796","created_date":"2025-10-10T00:00:00"}