{"id":"https://openalex.org/W7164549816","doi":"https://doi.org/10.1109/msp.2026.3665010","title":"On the Convergence, Implicit Bias, and Edge of Stability of Gradient Descent in Deep Learning: Reviewing recent progress [Special Issue on the Mathematics of Deep Learning]","display_name":"On the Convergence, Implicit Bias, and Edge of Stability of Gradient Descent in Deep Learning: Reviewing recent progress [Special Issue on the Mathematics of Deep Learning]","publication_year":2026,"publication_date":"2026-05-01","ids":{"openalex":"https://openalex.org/W7164549816","doi":"https://doi.org/10.1109/msp.2026.3665010"},"language":null,"primary_location":{"id":"doi:10.1109/msp.2026.3665010","is_oa":false,"landing_page_url":"https://doi.org/10.1109/msp.2026.3665010","pdf_url":null,"source":{"id":"https://openalex.org/S120977877","display_name":"IEEE Signal Processing Magazine","issn_l":"1053-5888","issn":["1053-5888","1558-0792"],"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 Signal Processing Magazine","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/A5038819585","display_name":"Hancheng Min","orcid":"https://orcid.org/0000-0003-0600-3854"},"institutions":[{"id":"https://openalex.org/I4210158445","display_name":"Institute of Natural Science","ror":"https://ror.org/004tze884","country_code":"KP","type":"education","lineage":["https://openalex.org/I4210158445"]}],"countries":["KP"],"is_corresponding":false,"raw_author_name":"Hancheng Min","raw_affiliation_strings":["Institute of Natural Sciences (INS), the School of Mathematical Sciences (SMS), and the MOE-LSC, Shanghai Jiao Tong University, Shanghai, China"],"raw_orcid":"https://orcid.org/0000-0003-0600-3854","affiliations":[{"raw_affiliation_string":"Institute of Natural Sciences (INS), the School of Mathematical Sciences (SMS), and the MOE-LSC, Shanghai Jiao Tong University, Shanghai, China","institution_ids":["https://openalex.org/I4210158445"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5065953740","display_name":"Lachlan Ewen MacDonald","orcid":"https://orcid.org/0000-0003-3601-9777"},"institutions":[{"id":"https://openalex.org/I4210113521","display_name":"Systems, Applications & Products in Data Processing (United States)","ror":"https://ror.org/01gbhv690","country_code":"US","type":"company","lineage":["https://openalex.org/I4210113521","https://openalex.org/I4210132444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Lachlan Ewen MacDonald","raw_affiliation_strings":["Center for Innovation in Data Engineering and Science (IDEAS) and the Department of Electrical and Systems Engineering (ESE), School of Engineering and Applied Science, University of Pennsylvania, Philadelphia, PA, USA"],"raw_orcid":"https://orcid.org/0000-0003-3601-9777","affiliations":[{"raw_affiliation_string":"Center for Innovation in Data Engineering and Science (IDEAS) and the Department of Electrical and Systems Engineering (ESE), School of Engineering and Applied Science, University of Pennsylvania, Philadelphia, PA, USA","institution_ids":["https://openalex.org/I4210113521"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5011256828","display_name":"Ren\u00e9 Vidal","orcid":"https://orcid.org/0000-0003-1838-0761"},"institutions":[{"id":"https://openalex.org/I79576946","display_name":"University of Pennsylvania","ror":"https://ror.org/00b30xv10","country_code":"US","type":"education","lineage":["https://openalex.org/I79576946"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ren\u00e9 Vidal","raw_affiliation_strings":["Department of Electrical and Systems Engineering and Department of Radiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Electrical and Systems Engineering and Department of Radiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA","institution_ids":["https://openalex.org/I79576946"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.95070972,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":null,"biblio":{"volume":"43","issue":"3","first_page":"37","last_page":"51"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.8853999972343445,"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":0.8853999972343445,"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/T10500","display_name":"Sparse and Compressive Sensing Techniques","score":0.020600000396370888,"subfield":{"id":"https://openalex.org/subfields/2206","display_name":"Computational Mechanics"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11206","display_name":"Model Reduction and Neural Networks","score":0.005400000140070915,"subfield":{"id":"https://openalex.org/subfields/3109","display_name":"Statistical and Nonlinear Physics"},"field":{"id":"https://openalex.org/fields/31","display_name":"Physics and Astronomy"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/stability","display_name":"Stability (learning theory)","score":0.6680999994277954},{"id":"https://openalex.org/keywords/enhanced-data-rates-for-gsm-evolution","display_name":"Enhanced Data Rates for GSM Evolution","score":0.5249000191688538},{"id":"https://openalex.org/keywords/gradient-descent","display_name":"Gradient descent","score":0.5210000276565552},{"id":"https://openalex.org/keywords/descent","display_name":"Descent (aeronautics)","score":0.41679999232292175},{"id":"https://openalex.org/keywords/calculus","display_name":"Calculus (dental)","score":0.3788999915122986},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.3630000054836273}],"concepts":[{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.6680999994277954},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.5249000191688538},{"id":"https://openalex.org/C153258448","wikidata":"https://www.wikidata.org/wiki/Q1199743","display_name":"Gradient descent","level":3,"score":0.5210000276565552},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.49470001459121704},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.4668999910354614},{"id":"https://openalex.org/C2776637919","wikidata":"https://www.wikidata.org/wiki/Q624380","display_name":"Descent (aeronautics)","level":2,"score":0.41679999232292175},{"id":"https://openalex.org/C2777686260","wikidata":"https://www.wikidata.org/wiki/Q144037","display_name":"Calculus (dental)","level":2,"score":0.3788999915122986},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.36899998784065247},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.3630000054836273},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3179999887943268},{"id":"https://openalex.org/C2988922011","wikidata":"https://www.wikidata.org/wiki/Q5449244","display_name":"Filtering theory","level":2,"score":0.3093999922275543},{"id":"https://openalex.org/C176321772","wikidata":"https://www.wikidata.org/wiki/Q1430640","display_name":"Numerical stability","level":3,"score":0.28940001130104065},{"id":"https://openalex.org/C136119220","wikidata":"https://www.wikidata.org/wiki/Q1000660","display_name":"Algebra over a field","level":2,"score":0.2863999903202057},{"id":"https://openalex.org/C206688291","wikidata":"https://www.wikidata.org/wiki/Q7617819","display_name":"Stochastic gradient descent","level":3,"score":0.28139999508857727},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2791999876499176},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.26350000500679016}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/msp.2026.3665010","is_oa":false,"landing_page_url":"https://doi.org/10.1109/msp.2026.3665010","pdf_url":null,"source":{"id":"https://openalex.org/S120977877","display_name":"IEEE Signal Processing Magazine","issn_l":"1053-5888","issn":["1053-5888","1558-0792"],"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 Signal Processing Magazine","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/10","score":0.5406320095062256,"display_name":"Reduced inequalities"}],"awards":[{"id":"https://openalex.org/G3020622031","display_name":null,"funder_award_id":"814201","funder_id":"https://openalex.org/F4320306164","funder_display_name":"Simons Foundation"}],"funders":[{"id":"https://openalex.org/F4320306164","display_name":"Simons Foundation","ror":"https://ror.org/01cmst727"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":8,"referenced_works":["https://openalex.org/W2566079294","https://openalex.org/W2746420172","https://openalex.org/W2810541051","https://openalex.org/W3025750359","https://openalex.org/W3127516423","https://openalex.org/W4206742934","https://openalex.org/W4250955649","https://openalex.org/W4377941395"],"related_works":[],"abstract_inverted_index":{"Deep":[0],"neural":[1],"networks":[2],"(DNNs)":[3],"trained":[4],"via":[5],"gradient":[6],"descent":[7],"(GD)":[8],"with":[9,157,188],"random":[10],"initialization":[11,98],"and":[12,57,94,169,182,203],"without":[13],"any":[14],"regularization":[15],"enjoy":[16],"good":[17],"generalization":[18,184],"performance":[19],"in":[20,153,166],"practice":[21],"despite":[22],"being":[23],"highly":[24],"overparametrized.":[25],"To":[26],"theoretically":[27,163],"understand":[28],"this":[29,50,67],"puzzling":[30],"phenomenon,":[31],"many":[32],"works":[33],"on":[34,40],"convergence":[35,112,168,181],"analysis":[36],"for":[37],"GD":[38,74,126,144],"algorithms":[39],"NNs":[41],"have":[42],"been":[43,131,150,195],"developed":[44],"over":[45],"the":[46,81,106,121,143,175],"last":[47],"half-decade.":[48],"In":[49],"article,":[51],"we":[52],"review":[53],"these":[54],"research":[55],"efforts":[56],"discuss":[58],"how":[59],"they":[60],"address":[61],"three":[62],"specific":[63],"questions":[64],"related":[65],"to":[66,100,174],"puzzle.":[68],"The":[69,116],"first":[70],"question":[71,118,177],"is":[72,119],"why":[73,120,179],"finds":[75],"a":[76,101,110,189,200],"global":[77,122],"minimum":[78,123],"efficiently,":[79],"which":[80,129,193],"literature":[82],"has":[83,130,149,194],"addressed":[84,132,197],"by":[85,125,133,198],"studying":[86],"what":[87,95],"level":[88],"of":[89,97,114,178],"overparametrization":[90,136],"(width,":[91],"depth,":[92],"etc.)":[93],"type":[96],"lead":[99],"benign":[102],"optimization":[103],"landscape":[104],"along":[105,142],"training":[107,155],"trajectory,":[108],"facilitating":[109],"linear":[111],"rate":[113],"GD.":[115],"next":[117],"found":[124],"generalizes":[127],"well,":[128],"showing":[134],"that":[135,152],"induces":[137],"an":[138],"implicit":[139,204],"simplicity":[140],"bias":[141,205],"trajectory.":[145],"More":[146],"recently,":[147],"it":[148],"observed":[151],"practice,":[154],"DNNs":[156],"far":[158],"larger":[159],"learning":[160,191],"rates":[161],"than":[162],"permissible":[164],"results":[165],"faster":[167,180],"better":[170,183],"generalization.":[171],"This":[172],"leads":[173],"third":[176],"can":[185],"be":[186],"achieved":[187],"large":[190],"rate,":[192],"recently":[196],"identifying":[199],"self-stabilization":[201],"mechanism":[202],"toward":[206],"flat":[207],"minima.":[208]},"counts_by_year":[],"updated_date":"2026-06-14T06:11:07.267592","created_date":"2026-06-13T00:00:00"}