{"id":"https://openalex.org/W7212007777","doi":"https://doi.org/10.48550/arxiv.2609.07868","title":"SGD in Multiclass Logistic Regression: Sequential Learning and Scaling Laws","display_name":"SGD in Multiclass Logistic Regression: Sequential Learning and Scaling Laws","publication_year":2026,"publication_date":"2026-09-07","ids":{"openalex":"https://openalex.org/W7212007777","doi":"https://doi.org/10.48550/arxiv.2609.07868"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.07868","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.07868","pdf_url":"https://arxiv.org/pdf/2609.07868","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2609.07868","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5055983865","display_name":"Konstantinos Tsiolis","orcid":"https://orcid.org/0000-0002-2654-0991","observed_orcids":["https://orcid.org/0000-0002-2654-0991"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tsiolis, Konstantinos Christopher","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5145679935","display_name":"Denny Wu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4387153999","display_name":"Flatiron Institute","ror":"https://ror.org/00sekdz59","country_code":"US","type":"nonprofit","lineage":["https://openalex.org/I4210107338","https://openalex.org/I4387153999"]},{"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":["US"],"is_corresponding":false,"raw_author_name":"Wu, Denny","raw_affiliation_strings":["New York University , Flatiron Institute"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"New York University , Flatiron Institute","institution_ids":["https://openalex.org/I4387153999","https://openalex.org/I57206974"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5024812488","display_name":"Christos Thrampoulidis","orcid":"https://orcid.org/0000-0001-9053-9365","observed_orcids":["https://orcid.org/0000-0001-9053-9365"]},"institutions":[{"id":"https://openalex.org/I141945490","display_name":"University of British Columbia","ror":"https://ror.org/03rmrcq20","country_code":"CA","type":"education","lineage":["https://openalex.org/I141945490"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Thrampoulidis, Christos","raw_affiliation_strings":["University of British Columbia"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of British Columbia","institution_ids":["https://openalex.org/I141945490"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5022491830","display_name":"Murat A. Erdogdu","orcid":"https://orcid.org/0000-0002-4271-8921","observed_orcids":["https://orcid.org/0000-0002-4271-8921"]},"institutions":[{"id":"https://openalex.org/I185261750","display_name":"University of Toronto","ror":"https://ror.org/03dbr7087","country_code":"CA","type":"education","lineage":["https://openalex.org/I185261750"]},{"id":"https://openalex.org/I4210127509","display_name":"Vector Institute","ror":"https://ror.org/03kqdja62","country_code":"CA","type":"facility","lineage":["https://openalex.org/I4210127509"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Erdogdu, Murat A.","raw_affiliation_strings":["University of Toronto , Vector Institute , Numurho"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Toronto , Vector Institute , Numurho","institution_ids":["https://openalex.org/I185261750","https://openalex.org/I4210127509"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":5,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9883000254631042,"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":0.9883000254631042,"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":0.003800000064074993,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.002300000051036477,"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/multiclass-logistic-regression","display_name":"multiclass logistic regression","score":0.9559999704360962},{"id":"https://openalex.org/keywords/stochastic-gradient-descent","display_name":"stochastic gradient descent","score":0.7440000176429749},{"id":"https://openalex.org/keywords/training-dynamics","display_name":"training dynamics","score":0.824999988079071},{"id":"https://openalex.org/keywords/scaling-laws","display_name":"scaling laws","score":0.9390000104904175},{"id":"https://openalex.org/keywords/gaussian-mixture-model","display_name":"Gaussian mixture model","score":0.8130000233650208},{"id":"https://openalex.org/keywords/sequential-learning","display_name":"sequential learning","score":0.9179999828338623},{"id":"https://openalex.org/keywords/model-capacity","display_name":"model capacity","score":0.8429999947547913},{"id":"https://openalex.org/keywords/high-dimensional-data-classification","display_name":"high-dimensional classification","score":0.6029999852180481}],"concepts":[{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.5799999833106995},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.45249998569488525},{"id":"https://openalex.org/C33676613","wikidata":"https://www.wikidata.org/wiki/Q13415176","display_name":"Dimension (graph theory)","level":2,"score":0.450300008058548},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.4262000024318695},{"id":"https://openalex.org/C87040749","wikidata":"https://www.wikidata.org/wiki/Q428971","display_name":"Power law","level":2,"score":0.4223000109195709},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4180999994277954},{"id":"https://openalex.org/C151956035","wikidata":"https://www.wikidata.org/wiki/Q1132755","display_name":"Logistic regression","level":2,"score":0.40709999203681946},{"id":"https://openalex.org/C163716315","wikidata":"https://www.wikidata.org/wiki/Q901177","display_name":"Gaussian","level":2,"score":0.4052000045776367},{"id":"https://openalex.org/C207390915","wikidata":"https://www.wikidata.org/wiki/Q1230525","display_name":"Divergence (linguistics)","level":2,"score":0.39590001106262207},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.3921000063419342},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.38109999895095825},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.3668000102043152},{"id":"https://openalex.org/C199539241","wikidata":"https://www.wikidata.org/wiki/Q7748","display_name":"Law","level":1,"score":0.3659999966621399},{"id":"https://openalex.org/C177769412","wikidata":"https://www.wikidata.org/wiki/Q278090","display_name":"Prior probability","level":3,"score":0.3546999990940094},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.35199999809265137},{"id":"https://openalex.org/C61326573","wikidata":"https://www.wikidata.org/wiki/Q1496376","display_name":"Gaussian process","level":3,"score":0.33820000290870667},{"id":"https://openalex.org/C123860398","wikidata":"https://www.wikidata.org/wiki/Q6934605","display_name":"Multiclass classification","level":3,"score":0.3271999955177307},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.31119999289512634},{"id":"https://openalex.org/C117568660","wikidata":"https://www.wikidata.org/wiki/Q1650843","display_name":"Multinomial logistic regression","level":2,"score":0.30239999294281006},{"id":"https://openalex.org/C57493831","wikidata":"https://www.wikidata.org/wiki/Q3134666","display_name":"Projection (relational algebra)","level":2,"score":0.2971000075340271},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.2782000005245209},{"id":"https://openalex.org/C180802074","wikidata":"https://www.wikidata.org/wiki/Q1052379","display_name":"Logistic function","level":2,"score":0.2743000090122223},{"id":"https://openalex.org/C48921125","wikidata":"https://www.wikidata.org/wiki/Q10861030","display_name":"Linear regression","level":2,"score":0.26600000262260437},{"id":"https://openalex.org/C205330730","wikidata":"https://www.wikidata.org/wiki/Q587182","display_name":"Logistic map","level":3,"score":0.2628999948501587},{"id":"https://openalex.org/C122123141","wikidata":"https://www.wikidata.org/wiki/Q176623","display_name":"Random variable","level":2,"score":0.2619999945163727},{"id":"https://openalex.org/C163175372","wikidata":"https://www.wikidata.org/wiki/Q3339222","display_name":"Linear model","level":2,"score":0.2567000091075897},{"id":"https://openalex.org/C83546350","wikidata":"https://www.wikidata.org/wiki/Q1139051","display_name":"Regression","level":2,"score":0.25609999895095825},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.25119999051094055},{"id":"https://openalex.org/C27931671","wikidata":"https://www.wikidata.org/wiki/Q7634497","display_name":"Sufficient dimension reduction","level":3,"score":0.25099998712539673}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.07868","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.07868","pdf_url":"https://arxiv.org/pdf/2609.07868","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2609.07868","doi":"https://doi.org/10.48550/arxiv.2609.07868","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.07868","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":"pmh:oai:arXiv.org:2609.07868","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.07868","pdf_url":"https://arxiv.org/pdf/2609.07868","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G4893171306","display_name":null,"funder_award_id":"ALLRP 581098-22","funder_id":"https://openalex.org/F4320334593","funder_display_name":"Natural Sciences and Engineering Research Council of Canada"},{"id":"https://openalex.org/G6290966371","display_name":null,"funder_award_id":"2019-06167","funder_id":"https://openalex.org/F4320334593","funder_display_name":"Natural Sciences and Engineering Research Council of Canada"}],"funders":[{"id":"https://openalex.org/F4320309949","display_name":"Canadian Institute for Advanced Research","ror":"https://ror.org/01sdtdd95"},{"id":"https://openalex.org/F4320319880","display_name":"Government of Canada","ror":"https://ror.org/010q4q527"},{"id":"https://openalex.org/F4320334593","display_name":"Natural Sciences and Engineering Research Council of Canada","ror":"https://ror.org/01h531d29"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7212007777.pdf","grobid_xml":"https://content.openalex.org/works/W7212007777.grobid-xml"},"referenced_works_count":14,"referenced_works":["https://openalex.org/W2775461895","https://openalex.org/W2790253170","https://openalex.org/W3001279689","https://openalex.org/W4307929880","https://openalex.org/W4414585696","https://openalex.org/W4416048255","https://openalex.org/W4416654762","https://openalex.org/W4417092181","https://openalex.org/W4417256404","https://openalex.org/W7092298369","https://openalex.org/W7127945923","https://openalex.org/W7128073161","https://openalex.org/W7133348504","https://openalex.org/W7143480655"],"related_works":["https://openalex.org/W7223069161","https://openalex.org/W2101153329","https://openalex.org/W3170039599","https://openalex.org/W3015944005","https://openalex.org/W7058043494","https://openalex.org/W4416048255","https://openalex.org/W2700343","https://openalex.org/W4416125235","https://openalex.org/W3009533209","https://openalex.org/W2112978490"],"abstract_inverted_index":{"We":[0,32,84],"study":[1],"the":[2,26,46,54,65,99,110,121],"training":[3,132,152],"dynamics":[4,56],"of":[5,18,156],"multiclass":[6,168],"logistic":[7,143],"regression":[8,166],"on":[9],"high-dimensional":[10],"Gaussian":[11],"mixture":[12],"models":[13],"with":[14,91,145],"a":[15,50,70,80,94,114,138],"large":[16],"number":[17],"classes":[19],"and":[20,79,124,151],"establish":[21],"precise":[22],"scaling":[23,140,162,174],"laws":[24,163,175],"governing":[25],"cross-entropy":[27],"risk":[28,55,111],"under":[29,93],"gradient-based":[30],"optimization.":[31],"show":[33],"that":[34],"learning":[35,77],"proceeds":[36],"sequentially":[37],"across":[38],"classes,":[39],"from":[40,164],"most":[41],"to":[42,167,172],"least":[43],"frequent.":[44],"When":[45,98],"class":[47,67],"priors":[48],"follow":[49],"power":[51,118,129],"law":[52,119,130,141],"distribution,":[53],"decompose":[57],"into":[58,113],"three":[59],"phases:":[60],"an":[61,125],"initial":[62],"plateau":[63],"until":[64],"first":[66],"is":[68,102],"learned,":[69],"power-law":[71],"decay":[72],"regime":[73],"during":[74],"which":[75],"sequential":[76],"occurs,":[78],"final":[81],"convergence":[82],"regime.":[83],"then":[85],"analyze":[86],"how":[87],"model":[88,149],"capacity":[89,115],"interacts":[90],"optimization":[92,126],"fixed":[95],"compute":[96],"budget.":[97],"effective":[100],"dimension":[101],"restricted":[103],"via":[104],"projection":[105],"onto":[106],"leading":[107],"principal":[108],"components,":[109],"decomposes":[112],"term":[116,127],"(a":[117,128],"in":[120,131,177],"retained":[122],"dimension)":[123],"time).":[133],"Optimizing":[134],"this":[135],"tradeoff":[136],"yields":[137],"compute-optimal":[139],"for":[142,148],"regression,":[144],"explicit":[146],"prescriptions":[147],"size":[150],"time":[153],"as":[154],"functions":[155],"compute.":[157],"These":[158],"results":[159],"extend":[160],"theoretical":[161],"linear":[165],"classification,":[169],"while":[170],"connecting":[171],"empirical":[173],"observed":[176],"large-scale":[178],"neural":[179],"networks.":[180]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-09-10T00:00:00"}