{"id":"https://openalex.org/W4407938032","doi":"https://doi.org/10.1109/tit.2025.3545564","title":"The Convex Landscape of Neural Networks: Characterizing Global Optima and Stationary Points via Lasso Models","display_name":"The Convex Landscape of Neural Networks: Characterizing Global Optima and Stationary Points via Lasso Models","publication_year":2025,"publication_date":"2025-02-25","ids":{"openalex":"https://openalex.org/W4407938032","doi":"https://doi.org/10.1109/tit.2025.3545564"},"language":"en","primary_location":{"id":"doi:10.1109/tit.2025.3545564","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tit.2025.3545564","pdf_url":null,"source":{"id":"https://openalex.org/S4502562","display_name":"IEEE Transactions on Information Theory","issn_l":"0018-9448","issn":["0018-9448","1557-9654"],"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 Information Theory","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5040173784","display_name":"Tolga Ergen","orcid":"https://orcid.org/0000-0003-4806-0224"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tolga Ergen","raw_affiliation_strings":["LG AI Research, Ann Arbor, MI, USA"],"raw_orcid":"https://orcid.org/0000-0003-4806-0224","affiliations":[{"raw_affiliation_string":"LG AI Research, Ann Arbor, MI, USA","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5001436196","display_name":"Mert Pilanc\u0131","orcid":"https://orcid.org/0000-0002-0870-9992"},"institutions":[{"id":"https://openalex.org/I97018004","display_name":"Stanford University","ror":"https://ror.org/00f54p054","country_code":"US","type":"education","lineage":["https://openalex.org/I97018004"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Mert Pilanci","raw_affiliation_strings":["Electrical Engineering Department, Stanford University, Stanford, CA, USA"],"raw_orcid":"https://orcid.org/0000-0002-0870-9992","affiliations":[{"raw_affiliation_string":"Electrical Engineering Department, Stanford University, Stanford, CA, USA","institution_ids":["https://openalex.org/I97018004"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":5.584,"has_fulltext":false,"cited_by_count":4,"citation_normalized_percentile":{"value":0.95109501,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":95,"max":98},"biblio":{"volume":"71","issue":"5","first_page":"3854","last_page":"3870"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10462","display_name":"Reinforcement Learning in Robotics","score":0.9732000231742859,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.9732000231742859,"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/T12056","display_name":"Markov Chains and Monte Carlo Methods","score":0.9663000106811523,"subfield":{"id":"https://openalex.org/subfields/2613","display_name":"Statistics and Probability"},"field":{"id":"https://openalex.org/fields/26","display_name":"Mathematics"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9599999785423279,"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/lasso","display_name":"Lasso (programming language)","score":0.6670480966567993},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.520913302898407},{"id":"https://openalex.org/keywords/regular-polygon","display_name":"Regular polygon","score":0.5105022192001343},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.47865140438079834},{"id":"https://openalex.org/keywords/mathematical-optimization","display_name":"Mathematical optimization","score":0.4600299596786499},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.34732067584991455},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.33311498165130615}],"concepts":[{"id":"https://openalex.org/C37616216","wikidata":"https://www.wikidata.org/wiki/Q3218363","display_name":"Lasso (programming language)","level":2,"score":0.6670480966567993},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.520913302898407},{"id":"https://openalex.org/C112680207","wikidata":"https://www.wikidata.org/wiki/Q714886","display_name":"Regular polygon","level":2,"score":0.5105022192001343},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.47865140438079834},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.4600299596786499},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.34732067584991455},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.33311498165130615},{"id":"https://openalex.org/C136764020","wikidata":"https://www.wikidata.org/wiki/Q466","display_name":"World Wide Web","level":1,"score":0.0},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tit.2025.3545564","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tit.2025.3545564","pdf_url":null,"source":{"id":"https://openalex.org/S4502562","display_name":"IEEE Transactions on Information Theory","issn_l":"0018-9448","issn":["0018-9448","1557-9654"],"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 Information Theory","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Climate action","score":0.5400000214576721,"id":"https://metadata.un.org/sdg/13"}],"awards":[{"id":"https://openalex.org/G1899085153","display_name":null,"funder_award_id":"W911NF-21-1-0242","funder_id":"https://openalex.org/F4320338281","funder_display_name":"Army Research Office"},{"id":"https://openalex.org/G4864293789","display_name":null,"funder_award_id":"ECCS-2037304","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G5346705942","display_name":null,"funder_award_id":"DMS-2134248","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G5998070949","display_name":null,"funder_award_id":"N00014-24-1-2164","funder_id":"https://openalex.org/F4320337345","funder_display_name":"Office of Naval Research"},{"id":"https://openalex.org/G8125586604","display_name":null,"funder_award_id":"CCF-2236829","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320337345","display_name":"Office of Naval Research","ror":"https://ror.org/00rk2pe57"},{"id":"https://openalex.org/F4320338281","display_name":"Army Research Office","ror":"https://ror.org/05epdh915"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":54,"referenced_works":["https://openalex.org/W180242331","https://openalex.org/W1986931325","https://openalex.org/W2000985550","https://openalex.org/W2030996579","https://openalex.org/W2034260606","https://openalex.org/W2116148865","https://openalex.org/W2118550318","https://openalex.org/W2120934134","https://openalex.org/W2129131372","https://openalex.org/W2135046866","https://openalex.org/W2138019504","https://openalex.org/W2161278885","https://openalex.org/W2194775991","https://openalex.org/W2608666998","https://openalex.org/W2785447469","https://openalex.org/W2894812324","https://openalex.org/W2964047251","https://openalex.org/W2983999360","https://openalex.org/W3044709444","https://openalex.org/W3113890314","https://openalex.org/W3160828921","https://openalex.org/W3162775926","https://openalex.org/W4235405218","https://openalex.org/W4250589301","https://openalex.org/W4250955649","https://openalex.org/W6633044140","https://openalex.org/W6676247742","https://openalex.org/W6677919164","https://openalex.org/W6684426766","https://openalex.org/W6729999211","https://openalex.org/W6730349232","https://openalex.org/W6743688258","https://openalex.org/W6752495264","https://openalex.org/W6752591435","https://openalex.org/W6758814661","https://openalex.org/W6763624801","https://openalex.org/W6764007832","https://openalex.org/W6764442487","https://openalex.org/W6768106036","https://openalex.org/W6773952175","https://openalex.org/W6774075890","https://openalex.org/W6774309501","https://openalex.org/W6777833935","https://openalex.org/W6779880336","https://openalex.org/W6780539426","https://openalex.org/W6783425961","https://openalex.org/W6783910908","https://openalex.org/W6787872581","https://openalex.org/W6791517735","https://openalex.org/W6796392226","https://openalex.org/W6797680811","https://openalex.org/W6797790463","https://openalex.org/W6838450934","https://openalex.org/W6846693424"],"related_works":["https://openalex.org/W4256576576","https://openalex.org/W3090384609","https://openalex.org/W3194013178","https://openalex.org/W4249086404","https://openalex.org/W2790019364","https://openalex.org/W2486360182","https://openalex.org/W2212119398","https://openalex.org/W4288780803","https://openalex.org/W2187089732","https://openalex.org/W3163178158"],"abstract_inverted_index":{"Due":[0,254],"to":[1,32,59,121,148,184,245,255,294],"the":[2,16,49,61,145,150,156,165,168,175,213,217,225,256,273],"non-convex":[3,19,252,288],"nature":[4],"of":[5,18,51,64,72,112,158,162,167,187,216,228],"training":[6,25,62,79,185,219],"Deep":[7],"Neural":[8],"Network":[9],"(DNN)":[10],"models,":[11],"their":[12,73],"effectiveness":[13],"relies":[14],"on":[15,78,198,272],"use":[17,50],"optimization":[20,53],"heuristics.":[21],"Traditional":[22],"methods":[23,31,289],"for":[24,125,299],"DNNs":[26],"often":[27],"require":[28],"costly":[29],"empirical":[30],"produce":[33],"successful":[34],"models":[35,58,235,284],"and":[36,55,67,87,130,160,265,290],"do":[37],"not":[38,292],"have":[39,269],"a":[40,69,95,102,110,192,204,229],"clear":[41],"theoretical":[42],"foundation.":[43],"In":[44],"this":[45],"study,":[46],"we":[47,132,190,280],"examine":[48],"convex":[52,97,140,151,231,234,241,283],"theory":[54,120],"sparse":[56,141],"recovery":[57,142],"refine":[60],"process":[63],"neural":[65,81,128],"networks":[66,82,129],"provide":[68],"better":[70],"interpretation":[71],"optimal":[74],"weights.":[75],"We":[76,115,208],"focus":[77],"two-layer":[80],"with":[83,203],"piecewise":[84],"linear":[85],"activations":[86],"demonstrate":[88],"that":[89,105,201,211,282],"they":[90],"can":[91,221,236,285],"be":[92,222,237],"formulated":[93],"as":[94,137,224,261],"finite-dimensional":[96],"program.":[98,232],"These":[99],"programs":[100],"include":[101],"regularization":[103],"term":[104],"promotes":[106],"sparsity,":[107],"which":[108,173],"constitutes":[109],"variant":[111],"group":[113],"Lasso.":[114],"first":[116],"utilize":[117],"semi-infinite":[118],"programming":[119],"prove":[122],"strong":[123],"duality":[124],"finite":[126],"width":[127],"then":[131],"express":[133],"these":[134],"architectures":[135],"equivalently":[136],"high":[138],"dimensional":[139],"models.":[143],"Remarkably,":[144],"worst-case":[146],"complexity":[147],"solve":[149],"program":[152],"is":[153,171,174,302],"polynomial":[154],"in":[155,177],"number":[157,161],"samples":[159],"neurons":[163],"when":[164],"rank":[166],"data":[169,186],"matrix":[170],"bounded,":[172],"case":[176],"convolutional":[178],"networks.":[179],"To":[180],"extend":[181],"our":[182,300],"method":[183],"arbitrary":[188],"rank,":[189],"develop":[191],"novel":[193],"polynomial-time":[194],"approximation":[195,206],"scheme":[196],"based":[197],"zonotope":[199],"subsampling":[200],"comes":[202],"guaranteed":[205],"ratio.":[207],"also":[209],"show":[210,281],"all":[212],"stationary":[214],"points":[215],"nonconvex":[218],"objective":[220],"characterized":[223],"global":[226],"optimum":[227],"subsampled":[230],"Our":[233],"trained":[238],"using":[239],"standard":[240],"solvers":[242],"without":[243],"resorting":[244],"heuristics":[246],"or":[247],"extensive":[248,277],"hyper-parameter":[249],"tuning":[250],"unlike":[251],"methods.":[253],"convexity,":[257],"optimizer":[258,295],"hyperparameters":[259],"such":[260],"initialization,":[262],"batch":[263],"sizes,":[264],"step":[266],"size":[267],"schedules":[268],"no":[270],"effect":[271],"final":[274],"model.":[275],"Through":[276],"numerical":[278],"experiments,":[279],"outperform":[286],"traditional":[287],"are":[291],"sensitive":[293],"hyperparameters.":[296],"The":[297],"code":[298],"experiments":[301],"available":[303],"at":[304],"<uri":[305],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[306],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">https://github.com/pilancilab/convex_nn</uri>.":[307]},"counts_by_year":[{"year":2026,"cited_by_count":2},{"year":2025,"cited_by_count":2}],"updated_date":"2025-12-19T19:40:27.379048","created_date":"2025-10-10T00:00:00"}