{"id":"https://openalex.org/W4412462829","doi":"https://doi.org/10.1088/2632-2153/adf02e","title":"Architecture-aware minimization (A2M): how to find flat minima in neural architecture search","display_name":"Architecture-aware minimization (A2M): how to find flat minima in neural architecture search","publication_year":2025,"publication_date":"2025-07-15","ids":{"openalex":"https://openalex.org/W4412462829","doi":"https://doi.org/10.1088/2632-2153/adf02e"},"language":"en","primary_location":{"id":"doi:10.1088/2632-2153/adf02e","doi":"https://doi.org/10.1088/2632-2153/adf02e","is_oa":true,"landing_page_url":"https://doi.org/10.1088/2632-2153/adf02e","pdf_url":null,"source":{"id":"https://openalex.org/S4210200687","display_name":"Machine Learning Science and Technology","issn_l":"2632-2153","issn":["2632-2153"],"is_oa":true,"is_in_doaj":true,"is_core":true,"listed_in":["cwts-core","doaj","dongbi-c","jufo-1","ki-jl-1","norway-1","oarl","poland-20","russia-white-list-1"],"host_organization":"https://openalex.org/P4310320083","host_organization_name":"IOP Publishing","host_organization_lineage":["https://openalex.org/P4310320083","https://openalex.org/P4310311669"],"host_organization_lineage_names":["IOP Publishing","Institute of Physics"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Machine Learning: Science and Technology","raw_type":"journal-article"},"volume":null,"type":"article","indexed_in":["arxiv","crossref","datacite","doaj"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1088/2632-2153/adf02e","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5008809497","display_name":"Matteo Gambella","orcid":"https://orcid.org/0000-0001-9734-670X","observed_orcids":["https://orcid.org/0000-0001-9734-670X"]},"institutions":[{"id":"https://openalex.org/I93860229","display_name":"Politecnico di Milano","ror":"https://ror.org/01nffqt88","country_code":"IT","type":"education","lineage":["https://openalex.org/I93860229"]}],"countries":["IT"],"is_corresponding":true,"raw_author_name":"Matteo Gambella","raw_affiliation_strings":["Department of Electronics, Information and Bioengineering, Politecnico di Milano, Via Ponzio 34/5, 20133 Milano, Italy"],"raw_orcid":"https://orcid.org/0000-0001-9734-670X","affiliations":[{"raw_affiliation_string":"Department of Electronics, Information and Bioengineering, Politecnico di Milano, Via Ponzio 34/5, 20133 Milano, Italy","institution_ids":["https://openalex.org/I93860229"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5035460406","display_name":"Fabrizio Pittorino","orcid":"https://orcid.org/0000-0002-1919-6141","observed_orcids":["https://orcid.org/0000-0002-1919-6141"]},"institutions":[{"id":"https://openalex.org/I93860229","display_name":"Politecnico di Milano","ror":"https://ror.org/01nffqt88","country_code":"IT","type":"education","lineage":["https://openalex.org/I93860229"]}],"countries":["IT"],"is_corresponding":true,"raw_author_name":"Fabrizio Pittorino","raw_affiliation_strings":["Department of Electronics, Information and Bioengineering, Politecnico di Milano, Via Ponzio 34/5, 20133 Milano, Italy"],"raw_orcid":"https://orcid.org/0000-0002-1919-6141","affiliations":[{"raw_affiliation_string":"Department of Electronics, Information and Bioengineering, Politecnico di Milano, Via Ponzio 34/5, 20133 Milano, Italy","institution_ids":["https://openalex.org/I93860229"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5035547226","display_name":"Manuel Roveri","orcid":"https://orcid.org/0000-0001-7828-7687","observed_orcids":["https://orcid.org/0000-0001-7828-7687"]},"institutions":[{"id":"https://openalex.org/I93860229","display_name":"Politecnico di Milano","ror":"https://ror.org/01nffqt88","country_code":"IT","type":"education","lineage":["https://openalex.org/I93860229"]}],"countries":["IT"],"is_corresponding":true,"raw_author_name":"Manuel Roveri","raw_affiliation_strings":["Department of Electronics, Information and Bioengineering, Politecnico di Milano, Via Ponzio 34/5, 20133 Milano, Italy"],"raw_orcid":"https://orcid.org/0000-0001-7828-7687","affiliations":[{"raw_affiliation_string":"Department of Electronics, Information and Bioengineering, Politecnico di Milano, Via Ponzio 34/5, 20133 Milano, Italy","institution_ids":["https://openalex.org/I93860229"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":["https://openalex.org/A5008809497","https://openalex.org/A5035460406","https://openalex.org/A5035547226"],"corresponding_institution_ids":["https://openalex.org/I93860229"],"apc_list":{"value":3000,"currency":"USD","value_usd":3000},"apc_paid":{"value":3000,"currency":"USD","value_usd":3000},"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.11164663,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"6","issue":"3","first_page":"035016","last_page":"035016"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.7544000148773193,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.7544000148773193,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.18330000340938568,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.013100000098347664,"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/neural-architecture-search","display_name":"neural architecture search","score":0.9929999709129333},{"id":"https://openalex.org/keywords/flat-minima","display_name":"flat minima","score":0.9169999957084656},{"id":"https://openalex.org/keywords/nas-bench-201","display_name":"NAS-Bench-201","score":0.8519999980926514},{"id":"https://openalex.org/keywords/double-asteroid-redirection-test","display_name":"DART","score":0.9629999995231628},{"id":"https://openalex.org/keywords/differentiable-architecture-search","display_name":"differentiable architecture search","score":0.8130000233650208},{"id":"https://openalex.org/keywords/loss-landscape","display_name":"loss landscape","score":0.4429999887943268},{"id":"https://openalex.org/keywords/automated-machine-learning","display_name":"automated machine learning","score":0.656000018119812}],"concepts":[{"id":"https://openalex.org/C186633575","wikidata":"https://www.wikidata.org/wiki/Q845060","display_name":"Maxima and minima","level":2,"score":0.8808062672615051},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.7149294018745422},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.7146254777908325},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.48736029863357544},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.47151386737823486},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.36963343620300293},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.32982751727104187},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.302610844373703},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.26064223051071167},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.17033013701438904},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.12232285737991333},{"id":"https://openalex.org/C166957645","wikidata":"https://www.wikidata.org/wiki/Q23498","display_name":"Archaeology","level":1,"score":0.0}],"mesh":[],"locations_count":5,"locations":[{"id":"doi:10.1088/2632-2153/adf02e","doi":"https://doi.org/10.1088/2632-2153/adf02e","is_oa":true,"landing_page_url":"https://doi.org/10.1088/2632-2153/adf02e","pdf_url":null,"source":{"id":"https://openalex.org/S4210200687","display_name":"Machine Learning Science and Technology","issn_l":"2632-2153","issn":["2632-2153"],"is_oa":true,"is_in_doaj":true,"is_core":true,"listed_in":["cwts-core","doaj","dongbi-c","jufo-1","ki-jl-1","norway-1","oarl","poland-20","russia-white-list-1"],"host_organization":"https://openalex.org/P4310320083","host_organization_name":"IOP Publishing","host_organization_lineage":["https://openalex.org/P4310320083","https://openalex.org/P4310311669"],"host_organization_lineage_names":["IOP Publishing","Institute of Physics"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Machine Learning: Science and Technology","raw_type":"journal-article"},{"id":"pmh:oai:arXiv.org:2503.10404","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2503.10404","pdf_url":"https://arxiv.org/pdf/2503.10404","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,"listed_in":[],"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"pmh:oai:doaj.org/article:6b995b4d4b4544db9238634adf6653df","doi":null,"is_oa":true,"landing_page_url":"https://doaj.org/article/6b995b4d4b4544db9238634adf6653df","pdf_url":null,"source":{"id":"https://openalex.org/S4306401280","display_name":"DOAJ (DOAJ: Directory of Open Access Journals)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-sa","license_id":"https://openalex.org/licenses/cc-by-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Machine Learning: Science and Technology, Vol 6, Iss 3, p 035016 (2025)","raw_type":"article"},{"id":"pmh:oai:re.public.polimi.it:11311/1295436","doi":null,"is_oa":true,"landing_page_url":"https://iopscience.iop.org/article/10.1088/2632-2153/adf02e","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,"listed_in":[],"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"info:eu-repo/semantics/article"},{"id":"doi:10.48550/arxiv.2503.10404","doi":"https://doi.org/10.48550/arxiv.2503.10404","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2503.10404","pdf_url":null,"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,"listed_in":[],"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.1088/2632-2153/adf02e","doi":"https://doi.org/10.1088/2632-2153/adf02e","is_oa":true,"landing_page_url":"https://doi.org/10.1088/2632-2153/adf02e","pdf_url":null,"source":{"id":"https://openalex.org/S4210200687","display_name":"Machine Learning Science and Technology","issn_l":"2632-2153","issn":["2632-2153"],"is_oa":true,"is_in_doaj":true,"is_core":true,"listed_in":["cwts-core","doaj","dongbi-c","jufo-1","ki-jl-1","norway-1","oarl","poland-20","russia-white-list-1"],"host_organization":"https://openalex.org/P4310320083","host_organization_name":"IOP Publishing","host_organization_lineage":["https://openalex.org/P4310320083","https://openalex.org/P4310311669"],"host_organization_lineage_names":["IOP Publishing","Institute of Physics"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Machine Learning: Science and Technology","raw_type":"journal-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412462829.pdf","grobid_xml":"https://content.openalex.org/works/W4412462829.grobid-xml"},"referenced_works_count":42,"referenced_works":["https://openalex.org/W2741430497","https://openalex.org/W2810075754","https://openalex.org/W2969797940","https://openalex.org/W2972830778","https://openalex.org/W2973943791","https://openalex.org/W2981613165","https://openalex.org/W2983785293","https://openalex.org/W2992043355","https://openalex.org/W2996409713","https://openalex.org/W2998438489","https://openalex.org/W3006500080","https://openalex.org/W3034334241","https://openalex.org/W3035620341","https://openalex.org/W3091401866","https://openalex.org/W3107893198","https://openalex.org/W3118344311","https://openalex.org/W3120629159","https://openalex.org/W3129626847","https://openalex.org/W3169943769","https://openalex.org/W3187567638","https://openalex.org/W3208681360","https://openalex.org/W4214717370","https://openalex.org/W4225905778","https://openalex.org/W4309374366","https://openalex.org/W4312545227","https://openalex.org/W4320233696","https://openalex.org/W4320858880","https://openalex.org/W4367626110","https://openalex.org/W4380558530","https://openalex.org/W4385572933","https://openalex.org/W4386573694","https://openalex.org/W4387910965","https://openalex.org/W4388116480","https://openalex.org/W4389130403","https://openalex.org/W4393064138","https://openalex.org/W4393153991","https://openalex.org/W4396926906","https://openalex.org/W4402352334","https://openalex.org/W4411244509","https://openalex.org/W4414693589","https://openalex.org/W6771859737","https://openalex.org/W6787972765"],"related_works":["https://openalex.org/W4411244509","https://openalex.org/W7127177558","https://openalex.org/W3156460968","https://openalex.org/W7203698759","https://openalex.org/W3016716171","https://openalex.org/W2924820442","https://openalex.org/W2995834909","https://openalex.org/W4411360370","https://openalex.org/W3036091446","https://openalex.org/W4386983325"],"abstract_inverted_index":{"Abstract":[0],"Neural":[1],"architecture":[2,27,39,48,87,153],"search":[3,40,88,176],"(NAS)":[4],"has":[5],"become":[6],"an":[7],"essential":[8],"tool":[9,226],"for":[10,139,227],"designing":[11],"effective":[12],"and":[13,37,53,61,169,174,201,230],"efficient":[14],"neural":[15,26,70],"networks.":[16],"In":[17,77],"this":[18],"paper,":[19],"we":[20,58,79,125],"investigate":[21],"the":[22,66,81,86,92,119,140,143,184],"geometric":[23],"properties":[24,68],"of":[25,45,69,85,94,118,145],"spaces":[28],"commonly":[29],"used":[30],"in":[31,47,74,106,152,232],"differentiable":[32,38,146,191,220],"NAS":[33,147,192,221],"methods,":[34,193],"specifically":[35],"NAS-Bench-201":[36,173],"(DARTS).":[41],"By":[42],"introducing":[43],"notions":[44],"flatness":[46,62],"space":[49],"such":[50],"as":[51],"neighborhoods":[52],"accuracy":[54,211],"barriers":[55,95],"along":[56],"paths,":[57],"reveal":[59],"locality":[60],"characteristics":[63],"analogous":[64],"to":[65,182],"well-known":[67],"network":[71],"loss":[72],"landscapes":[73],"weight":[75],"space.":[76,154],"particular,":[78],"unveil":[80],"detailed":[82],"geometrical":[83],"structure":[84],"landscape":[89],"by":[90,194],"uncovering":[91],"absence":[93],"between":[96],"well-performing":[97],"architectures,":[98],"finding":[99,207],"that":[100,136],"highly":[101],"accurate":[102],"architectures":[103,111,208],"cluster":[104],"together":[105],"flat":[107,150],"regions,":[108],"while":[109,206],"suboptimal":[110],"instead":[112],"remain":[113],"isolated,":[114],"showing":[115],"higher":[116],"values":[117],"barriers.":[120,212],"Building":[121],"on":[122,163,187,196,199,203],"these":[123],"insights,":[124],"propose":[126],"architecture-aware":[127],"minimization":[128],"(A2M),":[129],"a":[130,224],"novel":[131],"analytically":[132],"derived":[133],"algorithmic":[134],"framework":[135],"explicitly":[137],"biases,":[138],"first":[141],"time,":[142],"gradient":[144],"methods":[148],"towards":[149],"minima":[151],"A2M":[155,179,213],"consistently":[156],"improves":[157],"generalization":[158],"over":[159],"state-of-the-art":[160],"DARTS-based":[161],"algorithms":[162],"benchmark":[164],"datasets":[165],"including":[166],"CIFAR-10,":[167,197],"CIFAR-100,":[168,200],"ImageNet-16-120,":[170],"across":[171,189],"both":[172],"DARTS":[175],"spaces.":[177],"Notably,":[178],"is":[180],"able":[181],"increase":[183],"test":[185],"accuracy,":[186],"average":[188],"different":[190],"+3.60%":[195],"+4.60%":[198],"+3.64%":[202],"ImageNet-16-120":[204],"-":[205],"with":[209],"low":[210],"can":[214],"be":[215],"easily":[216],"integrated":[217],"into":[218],"existing":[219],"frameworks,":[222],"offering":[223],"versatile":[225],"future":[228],"research":[229],"applications":[231],"automated":[233],"machine":[234],"learning.":[235],"We":[236],"will":[237],"open-source":[238],"our":[239],"code":[240],"at":[241],"https://github.com/AI-Tech-Research-Lab/AsquaredM":[242],".":[243]},"counts_by_year":[],"updated_date":"2026-11-04T10:54:05.112534","created_date":"2025-10-10T00:00:00"}