{"id":"https://openalex.org/W7221086375","doi":"https://doi.org/10.48550/arxiv.2610.11668","title":"$\u03c3$Transfer: Uncertainty Transfer from Small to Large Networks under $\u03bc\\mathrm{P}$","display_name":"$\u03c3$Transfer: Uncertainty Transfer from Small to Large Networks under $\u03bc\\mathrm{P}$","publication_year":2026,"publication_date":"2026-10-08","ids":{"openalex":"https://openalex.org/W7221086375","doi":"https://doi.org/10.48550/arxiv.2610.11668"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2610.11668","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.11668","pdf_url":"https://arxiv.org/pdf/2610.11668","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/2610.11668","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5092698766","display_name":"Richard Bergna","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I241749","display_name":"University of Cambridge","ror":"https://ror.org/013meh722","country_code":"GB","type":"education","lineage":["https://openalex.org/I241749"]},{"id":"https://openalex.org/I4401726855","display_name":"Spotify (Sweden)","ror":"https://ror.org/00hbd6420","country_code":"SE","type":"company","lineage":["https://openalex.org/I4401726855"]}],"countries":["GB","SE"],"is_corresponding":false,"raw_author_name":"Bergna, Richard","raw_affiliation_strings":["University of Cambridge","Spotify"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Cambridge","institution_ids":["https://openalex.org/I241749"]},{"raw_affiliation_string":"Spotify","institution_ids":["https://openalex.org/I4401726855"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5158130268","display_name":"Fernando Ruiz Mazo","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I241749","display_name":"University of Cambridge","ror":"https://ror.org/013meh722","country_code":"GB","type":"education","lineage":["https://openalex.org/I241749"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Mazo, Fernando Ruiz","raw_affiliation_strings":["University of Cambridge"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Cambridge","institution_ids":["https://openalex.org/I241749"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5047511363","display_name":"Nicol\u00f2 Felicioni","orcid":"https://orcid.org/0000-0002-3555-7760","observed_orcids":["https://orcid.org/0000-0002-3555-7760"]},"institutions":[{"id":"https://openalex.org/I4401726855","display_name":"Spotify (Sweden)","ror":"https://ror.org/00hbd6420","country_code":"SE","type":"company","lineage":["https://openalex.org/I4401726855"]}],"countries":["SE"],"is_corresponding":false,"raw_author_name":"Felicioni, Nicol\u00f2","raw_affiliation_strings":["Spotify"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Spotify","institution_ids":["https://openalex.org/I4401726855"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5089201450","display_name":"Jose Miguel Hernandez Lobato","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I241749","display_name":"University of Cambridge","ror":"https://ror.org/013meh722","country_code":"GB","type":"education","lineage":["https://openalex.org/I241749"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Hern\u00e1ndez-Lobato, Jos\u00e9 Miguel","raw_affiliation_strings":["University of Cambridge"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Cambridge","institution_ids":["https://openalex.org/I241749"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5043239539","display_name":"Kamil Ciosek","orcid":"https://orcid.org/0000-0002-0238-9393","observed_orcids":["https://orcid.org/0000-0002-0238-9393"]},"institutions":[{"id":"https://openalex.org/I4401726855","display_name":"Spotify (Sweden)","ror":"https://ror.org/00hbd6420","country_code":"SE","type":"company","lineage":["https://openalex.org/I4401726855"]}],"countries":["SE"],"is_corresponding":false,"raw_author_name":"Ciosek, Kamil","raw_affiliation_strings":["Spotify"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Spotify","institution_ids":["https://openalex.org/I4401726855"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"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/T12814","display_name":"Gaussian Processes and Bayesian Inference","score":0.9229000210762024,"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/T12814","display_name":"Gaussian Processes and Bayesian Inference","score":0.9229000210762024,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.053599998354911804,"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/T10320","display_name":"Neural Networks and Applications","score":0.002400000113993883,"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/uncertainty-transfer","display_name":"uncertainty transfer","score":0.5870000123977661},{"id":"https://openalex.org/keywords/laplace-approximation","display_name":"Laplace approximation","score":0.7559999823570251},{"id":"https://openalex.org/keywords/zero-shot-transfer","display_name":"zero-shot transfer","score":0.5870000123977661},{"id":"https://openalex.org/keywords/uncertainty-estimation","display_name":"uncertainty estimation","score":0.33899998664855957},{"id":"https://openalex.org/keywords/out-of-distribution-detection","display_name":"out-of-distribution detection","score":0.8920000195503235}],"concepts":[{"id":"https://openalex.org/C178650346","wikidata":"https://www.wikidata.org/wiki/Q201984","display_name":"Covariance","level":2,"score":0.6017000079154968},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.5210999846458435},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4846000075340271},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.48100000619888306},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.47769999504089355},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.4309000074863434},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.41510000824928284},{"id":"https://openalex.org/C202887219","wikidata":"https://www.wikidata.org/wiki/Q3895221","display_name":"Parametrization (atmospheric modeling)","level":3,"score":0.4050999879837036},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.4050000011920929},{"id":"https://openalex.org/C2776175482","wikidata":"https://www.wikidata.org/wiki/Q1195816","display_name":"Transfer (computing)","level":2,"score":0.3912000060081482},{"id":"https://openalex.org/C97937538","wikidata":"https://www.wikidata.org/wiki/Q199691","display_name":"Laplace transform","level":2,"score":0.38929998874664307},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.36629998683929443},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.3662000000476837},{"id":"https://openalex.org/C81299745","wikidata":"https://www.wikidata.org/wiki/Q334269","display_name":"Transfer function","level":2,"score":0.3262999951839447},{"id":"https://openalex.org/C81692654","wikidata":"https://www.wikidata.org/wiki/Q225926","display_name":"Kriging","level":2,"score":0.30880001187324524},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.30709999799728394},{"id":"https://openalex.org/C47446073","wikidata":"https://www.wikidata.org/wiki/Q5165890","display_name":"Control theory (sociology)","level":3,"score":0.29840001463890076},{"id":"https://openalex.org/C57830394","wikidata":"https://www.wikidata.org/wiki/Q278079","display_name":"Posterior probability","level":3,"score":0.2930000126361847},{"id":"https://openalex.org/C2984842247","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep neural networks","level":3,"score":0.27720001339912415},{"id":"https://openalex.org/C137209882","wikidata":"https://www.wikidata.org/wiki/Q1403517","display_name":"Measurement uncertainty","level":2,"score":0.26109999418258667},{"id":"https://openalex.org/C165646398","wikidata":"https://www.wikidata.org/wiki/Q3755281","display_name":"Minimum-variance unbiased estimator","level":3,"score":0.2590999901294708},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.2572999894618988}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2610.11668","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.11668","pdf_url":"https://arxiv.org/pdf/2610.11668","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.2610.11668","doi":"https://doi.org/10.48550/arxiv.2610.11668","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2610.11668","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2610.11668","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.11668","pdf_url":"https://arxiv.org/pdf/2610.11668","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":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7221086375.pdf","grobid_xml":"https://content.openalex.org/works/W7221086375.grobid-xml"},"referenced_works_count":9,"referenced_works":["https://openalex.org/W1872312298","https://openalex.org/W2531327146","https://openalex.org/W2750384547","https://openalex.org/W2902986194","https://openalex.org/W3038074040","https://openalex.org/W3088274487","https://openalex.org/W4385964757","https://openalex.org/W6907647838","https://openalex.org/W6945134980"],"related_works":["https://openalex.org/W7221125729","https://openalex.org/W6986697597","https://openalex.org/W4221164116","https://openalex.org/W3092612693","https://openalex.org/W7126281328","https://openalex.org/W7047471083","https://openalex.org/W7160926641","https://openalex.org/W4415798877","https://openalex.org/W7024026090","https://openalex.org/W2786857698"],"abstract_inverted_index":{"Reliable":[0],"predictive":[1],"uncertainty":[2],"in":[3],"Laplace":[4],"approximations":[5],"depends":[6],"critically":[7],"on":[8,62,81,129],"the":[9,31,41,46,60,71,79,82,91],"prior":[10,42,92],"precision,":[11,97],"yet":[12],"selecting":[13],"it":[14,69],"requires":[15],"a":[16,38,63,132,139,146,158,183],"posterior":[17,94,170],"sweep":[18],"that":[19,44],"is":[20],"prohibitively":[21],"expensive":[22],"for":[23,78],"neural":[24],"networks":[25],"with":[26,157],"billions":[27],"of":[28,40,90,135,150,175],"parameters.":[29],"Under":[30],"Maximal":[32],"Update":[33],"Parametrization":[34],"($\u03bc\\mathrm{P}$),":[35],"we":[36,58],"derive":[37],"rescaling":[39],"covariance":[43],"makes":[45],"selected":[47,96],"precision":[48,61,80],"stable":[49],"as":[50],"model":[51,65,84,144],"width":[52,125],"grows.":[53],"This":[54],"leads":[55],"to":[56,70,127,142,154],"$\u03c3\\mathrm{Transfer}$:":[57],"select":[59],"smaller":[64],"and":[66,98,104,111,178],"zero-shot":[67],"transfer":[68,174],"much":[72],"larger":[73,83],"model,":[74],"i.e.,":[75],"without":[76,181],"searching":[77],"at":[85,131],"all.":[86],"We":[87],"show":[88],"convergence":[89],"kernel,":[93],"covariance,":[95],"posterior-derived":[99],"decisions":[100,180],"under":[101],"explicit":[102],"conditions,":[103],"verify":[105],"$\u03c3\\mathrm{Transfer}$":[106],"across":[107,165],"regression,":[108],"image":[109],"classification,":[110],"Transformer":[112],"readouts.":[113],"For":[114],"example,":[115],"measured":[116,160],"precision-sweep":[117],"speedups":[118],"reach":[119],"$\\sim":[120,151,155],"5000\\times$":[121],"when":[122],"transferring":[123,137],"from":[124,138],"128":[126],"4096":[128],"MNIST,":[130],"target-NLL":[133,161],"degradation":[134],"$0.002$;":[136],"public":[140],"1B":[141],"7B":[143],"gives":[145],"median":[147],"search":[148],"speedup":[149],"2.3\\times$":[152],"(up":[153],"330\\times$),":[156],"mean":[159],"increase":[162],"below":[163],"$10^{-4}$":[164],"ten":[166],"tasks.":[167],"The":[168],"same":[169],"stability":[171],"also":[172],"enables":[173],"acquisition,":[176],"OOD-detection,":[177],"abstention":[179],"constructing":[182],"target":[184],"posterior.":[185]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-10-10T00:00:00"}