{"id":"https://openalex.org/W2998438489","doi":"https://doi.org/10.1073/pnas.1908636117","title":"Shaping the learning landscape in neural networks around wide flat minima","display_name":"Shaping the learning landscape in neural networks around wide flat minima","publication_year":2019,"publication_date":"2019-12-23","ids":{"openalex":"https://openalex.org/W2998438489","doi":"https://doi.org/10.1073/pnas.1908636117","mag":"2998438489","pmid":"https://pubmed.ncbi.nlm.nih.gov/31871189"},"language":"en","primary_location":{"id":"doi:10.1073/pnas.1908636117","is_oa":true,"landing_page_url":"https://doi.org/10.1073/pnas.1908636117","pdf_url":null,"source":{"id":"https://openalex.org/S125754415","display_name":"Proceedings of the National Academy of Sciences","issn_l":"0027-8424","issn":["0027-8424","1091-6490"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320052","host_organization_name":"National Academy of Sciences","host_organization_lineage":["https://openalex.org/P4310320052"],"host_organization_lineage_names":["National Academy of Sciences"],"type":"journal"},"license":"cc-by-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the National Academy of Sciences","raw_type":"journal-article"},"type":"article","indexed_in":["arxiv","crossref","pubmed"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://doi.org/10.1073/pnas.1908636117","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5028083013","display_name":"Carlo Baldassi","orcid":"https://orcid.org/0000-0002-5451-8388"},"institutions":[{"id":"https://openalex.org/I4210099850","display_name":"Istituto Nazionale di Fisica Nucleare, Sezione di Torino","ror":"https://ror.org/01vj6ck58","country_code":"IT","type":"facility","lineage":["https://openalex.org/I160013858","https://openalex.org/I4210099850"]},{"id":"https://openalex.org/I71209653","display_name":"Bocconi University","ror":"https://ror.org/05crjpb27","country_code":"IT","type":"education","lineage":["https://openalex.org/I71209653"]}],"countries":["IT"],"is_corresponding":true,"raw_author_name":"Carlo Baldassi","raw_affiliation_strings":["Artificial Intelligence Lab, Institute for Data Science and Analytics, Bocconi University, 20136 Milan, Italy;","Istituto Nazionale di Fisica Nucleare, Sezione di Torino, 10125 Torino, Italy;"],"raw_orcid":"https://orcid.org/0000-0002-5451-8388","affiliations":[{"raw_affiliation_string":"Artificial Intelligence Lab, Institute for Data Science and Analytics, Bocconi University, 20136 Milan, Italy;","institution_ids":["https://openalex.org/I71209653"]},{"raw_affiliation_string":"Istituto Nazionale di Fisica Nucleare, Sezione di Torino, 10125 Torino, Italy;","institution_ids":["https://openalex.org/I4210099850"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5035460406","display_name":"Fabrizio Pittorino","orcid":"https://orcid.org/0000-0002-1919-6141"},"institutions":[{"id":"https://openalex.org/I177477856","display_name":"Politecnico di Torino","ror":"https://ror.org/00bgk9508","country_code":"IT","type":"education","lineage":["https://openalex.org/I177477856"]},{"id":"https://openalex.org/I71209653","display_name":"Bocconi University","ror":"https://ror.org/05crjpb27","country_code":"IT","type":"education","lineage":["https://openalex.org/I71209653"]}],"countries":["IT"],"is_corresponding":false,"raw_author_name":"Fabrizio Pittorino","raw_affiliation_strings":["Artificial Intelligence Lab, Institute for Data Science and Analytics, Bocconi University, 20136 Milan, Italy;","Department of Applied Science and Technology, Politecnico di Torino, 10129 Torino, Italy;"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Artificial Intelligence Lab, Institute for Data Science and Analytics, Bocconi University, 20136 Milan, Italy;","institution_ids":["https://openalex.org/I71209653"]},{"raw_affiliation_string":"Department of Applied Science and Technology, Politecnico di Torino, 10129 Torino, Italy;","institution_ids":["https://openalex.org/I177477856"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5041685522","display_name":"Riccardo Zecchina","orcid":"https://orcid.org/0000-0002-1221-5207"},"institutions":[{"id":"https://openalex.org/I12323705","display_name":"The Abdus Salam International Centre for Theoretical Physics (ICTP)","ror":"https://ror.org/009gyvm78","country_code":"IT","type":"facility","lineage":["https://openalex.org/I12323705","https://openalex.org/I1286959531","https://openalex.org/I1293226324","https://openalex.org/I2801247003"]},{"id":"https://openalex.org/I71209653","display_name":"Bocconi University","ror":"https://ror.org/05crjpb27","country_code":"IT","type":"education","lineage":["https://openalex.org/I71209653"]}],"countries":["IT"],"is_corresponding":true,"raw_author_name":"Riccardo Zecchina","raw_affiliation_strings":["Artificial Intelligence Lab, Institute for Data Science and Analytics, Bocconi University, 20136 Milan, Italy;","International Centre for Theoretical Physics, 34151 Trieste, Italy"],"raw_orcid":"https://orcid.org/0000-0002-1221-5207","affiliations":[{"raw_affiliation_string":"Artificial Intelligence Lab, Institute for Data Science and Analytics, Bocconi University, 20136 Milan, Italy;","institution_ids":["https://openalex.org/I71209653"]},{"raw_affiliation_string":"International Centre for Theoretical Physics, 34151 Trieste, Italy","institution_ids":["https://openalex.org/I12323705"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5028083013","https://openalex.org/A5041685522"],"corresponding_institution_ids":["https://openalex.org/I12323705","https://openalex.org/I4210099850","https://openalex.org/I71209653"],"apc_list":null,"apc_paid":null,"fwci":6.2402,"has_fulltext":false,"cited_by_count":91,"citation_normalized_percentile":{"value":0.96989675,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":96,"max":100},"biblio":{"volume":"117","issue":"1","first_page":"161","last_page":"170"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9998999834060669,"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":0.9998999834060669,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9988999962806702,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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":0.9987000226974487,"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/maxima-and-minima","display_name":"Maxima and minima","score":0.9025156497955322},{"id":"https://openalex.org/keywords/overfitting","display_name":"Overfitting","score":0.8183611631393433},{"id":"https://openalex.org/keywords/hessian-matrix","display_name":"Hessian matrix","score":0.7523115873336792},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.6564522385597229},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.5544643402099609},{"id":"https://openalex.org/keywords/gradient-descent","display_name":"Gradient descent","score":0.48132479190826416},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4764552414417267},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.4638177752494812},{"id":"https://openalex.org/keywords/focus","display_name":"Focus (optics)","score":0.43949949741363525},{"id":"https://openalex.org/keywords/stochastic-gradient-descent","display_name":"Stochastic gradient descent","score":0.42138671875},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.39771613478660583},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.36261647939682007},{"id":"https://openalex.org/keywords/mathematical-optimization","display_name":"Mathematical optimization","score":0.34326887130737305},{"id":"https://openalex.org/keywords/applied-mathematics","display_name":"Applied mathematics","score":0.2285524308681488},{"id":"https://openalex.org/keywords/mathematical-analysis","display_name":"Mathematical analysis","score":0.1407325267791748},{"id":"https://openalex.org/keywords/physics","display_name":"Physics","score":0.10859698057174683}],"concepts":[{"id":"https://openalex.org/C186633575","wikidata":"https://www.wikidata.org/wiki/Q845060","display_name":"Maxima and minima","level":2,"score":0.9025156497955322},{"id":"https://openalex.org/C22019652","wikidata":"https://www.wikidata.org/wiki/Q331309","display_name":"Overfitting","level":3,"score":0.8183611631393433},{"id":"https://openalex.org/C203616005","wikidata":"https://www.wikidata.org/wiki/Q620495","display_name":"Hessian matrix","level":2,"score":0.7523115873336792},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.6564522385597229},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5544643402099609},{"id":"https://openalex.org/C153258448","wikidata":"https://www.wikidata.org/wiki/Q1199743","display_name":"Gradient descent","level":3,"score":0.48132479190826416},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4764552414417267},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.4638177752494812},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.43949949741363525},{"id":"https://openalex.org/C206688291","wikidata":"https://www.wikidata.org/wiki/Q7617819","display_name":"Stochastic gradient descent","level":3,"score":0.42138671875},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.39771613478660583},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.36261647939682007},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.34326887130737305},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.2285524308681488},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.1407325267791748},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.10859698057174683},{"id":"https://openalex.org/C120665830","wikidata":"https://www.wikidata.org/wiki/Q14620","display_name":"Optics","level":1,"score":0.0}],"mesh":[],"locations_count":5,"locations":[{"id":"doi:10.1073/pnas.1908636117","is_oa":true,"landing_page_url":"https://doi.org/10.1073/pnas.1908636117","pdf_url":null,"source":{"id":"https://openalex.org/S125754415","display_name":"Proceedings of the National Academy of Sciences","issn_l":"0027-8424","issn":["0027-8424","1091-6490"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320052","host_organization_name":"National Academy of Sciences","host_organization_lineage":["https://openalex.org/P4310320052"],"host_organization_lineage_names":["National Academy of Sciences"],"type":"journal"},"license":"cc-by-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the National Academy of Sciences","raw_type":"journal-article"},{"id":"pmid:31871189","is_oa":false,"landing_page_url":"https://pubmed.ncbi.nlm.nih.gov/31871189","pdf_url":null,"source":{"id":"https://openalex.org/S4306525036","display_name":"PubMed","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I1299303238","host_organization_name":"National Institutes of Health","host_organization_lineage":["https://openalex.org/I1299303238"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the National Academy of Sciences of the United States of America","raw_type":null},{"id":"pmh:oai:arXiv.org:1905.07833","is_oa":true,"landing_page_url":"http://arxiv.org/abs/1905.07833","pdf_url":"https://arxiv.org/pdf/1905.07833","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"},{"id":"pmh:oai:pubmedcentral.nih.gov:6955380","is_oa":true,"landing_page_url":"https://www.ncbi.nlm.nih.gov/pmc/articles/6955380","pdf_url":null,"source":{"id":"https://openalex.org/S2764455111","display_name":"PubMed Central","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I1299303238","host_organization_name":"National Institutes of Health","host_organization_lineage":["https://openalex.org/I1299303238"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Proc Natl Acad Sci U S A","raw_type":"Text"},{"id":"pmh:oai:re.public.polimi.it:11311/1252442","is_oa":true,"landing_page_url":"https://hdl.handle.net/11311/1252442","pdf_url":null,"source":{"id":"https://openalex.org/S4306400312","display_name":"Virtual Community of Pathological Anatomy (University of Castilla La Mancha)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I79189158","host_organization_name":"University of Castilla-La Mancha","host_organization_lineage":["https://openalex.org/I79189158"],"host_organization_lineage_names":[],"type":"repository"},"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"info:eu-repo/semantics/article"}],"best_oa_location":{"id":"doi:10.1073/pnas.1908636117","is_oa":true,"landing_page_url":"https://doi.org/10.1073/pnas.1908636117","pdf_url":null,"source":{"id":"https://openalex.org/S125754415","display_name":"Proceedings of the National Academy of Sciences","issn_l":"0027-8424","issn":["0027-8424","1091-6490"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320052","host_organization_name":"National Academy of Sciences","host_organization_lineage":["https://openalex.org/P4310320052"],"host_organization_lineage_names":["National Academy of Sciences"],"type":"journal"},"license":"cc-by-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the National Academy of Sciences","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G6607879451","display_name":null,"funder_award_id":"N00014-17-1-2569","funder_id":"https://openalex.org/F4320337345","funder_display_name":"Office of Naval Research"},{"id":"https://openalex.org/G7272902013","display_name":null,"funder_award_id":"N00014-17-1-2569","funder_id":"https://openalex.org/F4320332923","funder_display_name":"U.S. Navy"}],"funders":[{"id":"https://openalex.org/F4320332923","display_name":"U.S. Navy","ror":"https://ror.org/03ar0mv07"},{"id":"https://openalex.org/F4320337345","display_name":"Office of Naval Research","ror":"https://ror.org/00rk2pe57"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":34,"referenced_works":["https://openalex.org/W121410702","https://openalex.org/W1607198972","https://openalex.org/W1650706190","https://openalex.org/W1967364564","https://openalex.org/W1980906795","https://openalex.org/W2013128153","https://openalex.org/W2027368911","https://openalex.org/W2032781595","https://openalex.org/W2033606703","https://openalex.org/W2036759514","https://openalex.org/W2056517685","https://openalex.org/W2060719094","https://openalex.org/W2070784404","https://openalex.org/W2085138324","https://openalex.org/W2088522405","https://openalex.org/W2091574922","https://openalex.org/W2104731515","https://openalex.org/W2128552304","https://openalex.org/W2211979669","https://openalex.org/W2262347687","https://openalex.org/W2262732936","https://openalex.org/W2484405816","https://openalex.org/W2517648041","https://openalex.org/W2750384547","https://openalex.org/W2766309150","https://openalex.org/W2919115771","https://openalex.org/W2963698262","https://openalex.org/W2963959597","https://openalex.org/W2973471489","https://openalex.org/W3103710211","https://openalex.org/W3140968660","https://openalex.org/W3146637812","https://openalex.org/W4206502013","https://openalex.org/W6675904171"],"related_works":["https://openalex.org/W2153649672","https://openalex.org/W2355987247","https://openalex.org/W2115886818","https://openalex.org/W3177326532","https://openalex.org/W3118431319","https://openalex.org/W3143650729","https://openalex.org/W3004700632","https://openalex.org/W2466106906","https://openalex.org/W4297883503","https://openalex.org/W3177888558"],"abstract_inverted_index":{"Learning":[0],"in":[1,37,59],"deep":[2],"neural":[3,85],"networks":[4],"takes":[5],"place":[6],"by":[7,15,214],"minimizing":[8],"a":[9,16,69,94,193,215],"nonconvex":[10,81],"high-dimensional":[11],"loss":[12,112,138],"function,":[13],"typically":[14],"stochastic":[17],"gradient":[18],"descent":[19],"(SGD)":[20],"strategy.":[21],"The":[22],"learning":[23,153,204],"process":[24,205],"is":[25,68],"observed":[26],"to":[27,30,208],"be":[28,55],"able":[29],"find":[31],"good":[32],"minimizers":[33,43,134],"without":[34],"getting":[35],"stuck":[36],"local":[38],"critical":[39,127],"points":[40],"and":[41,71,83,92,99,126,169,187,229],"such":[42],"are":[44],"often":[45],"satisfactory":[46],"at":[47],"avoiding":[48],"overfitting.":[49],"How":[50],"these":[51],"2":[52],"features":[53,101],"can":[54],"kept":[56],"under":[57],"control":[58],"nonlinear":[60],"devices":[61,154],"composed":[62],"of":[63,65,96,135,144,152,180,185,196,199,218,224],"millions":[64],"tunable":[66],"connections":[67],"profound":[70],"far-reaching":[72],"open":[73],"question.":[74],"In":[75,182],"this":[76],"paper":[77],"we":[78,165,190],"study":[79,217],"basic":[80,97],"1-":[82],"2-layer":[84],"network":[86],"models":[87],"that":[88,109,132,172,192],"learn":[89],"random":[90],"patterns":[91],"derive":[93,166],"number":[95],"geometrical":[98],"algorithmic":[100,163],"which":[102,121,156],"suggest":[103],"some":[104],"answers.":[105],"We":[106,129,148,210],"first":[107],"show":[108,131,150,191],"the":[110,133,136,142,145,162,183,197,200,203,212,219,222,225],"error":[111,146],"function":[113,139],"presents":[114],"few":[115],"extremely":[116],"wide":[117,177],"flat":[118,178],"minima":[119,125],"(WFM)":[120],"coexist":[122],"with":[123,141],"narrower":[124],"points.":[128],"then":[130],"cross-entropy":[137,188],"overlap":[140],"WFM":[143,157],"loss.":[147],"also":[149,206],"examples":[151],"for":[155],"do":[158],"not":[159],"exist.":[160],"From":[161],"perspective":[164],"entropy-driven":[167],"greedy":[168],"message-passing":[170],"algorithms":[171],"focus":[173],"their":[174,227,230],"search":[175],"on":[176,233],"regions":[179],"minimizers.":[181],"case":[184],"SGD":[186],"loss,":[189],"slow":[194],"reduction":[195],"norm":[198],"weights":[201],"along":[202],"leads":[207],"WFM.":[209],"corroborate":[211],"results":[213],"numerical":[216],"correlations":[220],"between":[221],"volumes":[223],"minimizers,":[226],"Hessian,":[228],"generalization":[231],"performance":[232],"real":[234],"data.":[235]},"counts_by_year":[{"year":2026,"cited_by_count":8},{"year":2025,"cited_by_count":14},{"year":2024,"cited_by_count":13},{"year":2023,"cited_by_count":16},{"year":2022,"cited_by_count":14},{"year":2021,"cited_by_count":14},{"year":2020,"cited_by_count":9},{"year":2019,"cited_by_count":3}],"updated_date":"2026-08-18T07:49:30.821534","created_date":"2025-10-10T00:00:00"}