{"id":"https://openalex.org/W7214799486","doi":"https://doi.org/10.48550/arxiv.2609.33835","title":"An Active-Bottleneck Mechanism for Weak-to-Strong Generalization","display_name":"An Active-Bottleneck Mechanism for Weak-to-Strong Generalization","publication_year":2026,"publication_date":"2026-09-27","ids":{"openalex":"https://openalex.org/W7214799486","doi":"https://doi.org/10.48550/arxiv.2609.33835"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.33835","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.33835","pdf_url":"https://arxiv.org/pdf/2609.33835","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/2609.33835","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5153491466","display_name":"Mohammad Zeinalpour","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I133529467","display_name":"Sharif University of Technology","ror":"https://ror.org/024c2fq17","country_code":"IR","type":"education","lineage":["https://openalex.org/I133529467"]}],"countries":["IR"],"is_corresponding":false,"raw_author_name":"Zeinalpour, Mohammad","raw_affiliation_strings":["Department of Computer Engineering , Sharif University of Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Engineering , Sharif University of Technology","institution_ids":["https://openalex.org/I133529467"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5018408976","display_name":"Amir Najafi","orcid":"https://orcid.org/0000-0002-6680-0110","observed_orcids":["https://orcid.org/0000-0002-6680-0110"]},"institutions":[{"id":"https://openalex.org/I133529467","display_name":"Sharif University of Technology","ror":"https://ror.org/024c2fq17","country_code":"IR","type":"education","lineage":["https://openalex.org/I133529467"]}],"countries":["IR"],"is_corresponding":false,"raw_author_name":"Najafi, Amir","raw_affiliation_strings":["Department of Computer Engineering , Sharif University of Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computer Engineering , Sharif University of Technology","institution_ids":["https://openalex.org/I133529467"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I133529467"],"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/T12072","display_name":"Machine Learning and Algorithms","score":0.5590999722480774,"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/T12072","display_name":"Machine Learning and Algorithms","score":0.5590999722480774,"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":0.22130000591278076,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.15860000252723694,"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/weak-to-strong-generalization","display_name":"weak-to-strong generalization","score":0.9980000257492065},{"id":"https://openalex.org/keywords/teacher-student-models","display_name":"teacher-student models","score":0.6420000195503235},{"id":"https://openalex.org/keywords/two-stage-learning","display_name":"two-stage learning","score":0.6970000267028809},{"id":"https://openalex.org/keywords/pseudo-labeling","display_name":"pseudo-labeling","score":0.8309999704360962},{"id":"https://openalex.org/keywords/spectral-decay","display_name":"spectral decay","score":0.875},{"id":"https://openalex.org/keywords/random-feature-models","display_name":"random feature models","score":0.6930000185966492},{"id":"https://openalex.org/keywords/unlabeled-data","display_name":"unlabeled data","score":0.4659999907016754}],"concepts":[{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.7602999806404114},{"id":"https://openalex.org/C45340560","wikidata":"https://www.wikidata.org/wiki/Q215382","display_name":"Disjoint sets","level":2,"score":0.6984000205993652},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.616599977016449},{"id":"https://openalex.org/C34388435","wikidata":"https://www.wikidata.org/wiki/Q2267362","display_name":"Bounded function","level":2,"score":0.6008999943733215},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.5598000288009644},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.5385000109672546},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.5149000287055969},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.48840001225471497},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.3869999945163727},{"id":"https://openalex.org/C117765406","wikidata":"https://www.wikidata.org/wiki/Q5362437","display_name":"Generalization error","level":3,"score":0.3628000020980835},{"id":"https://openalex.org/C202444582","wikidata":"https://www.wikidata.org/wiki/Q837863","display_name":"Pure mathematics","level":1,"score":0.3499000072479248},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.34360000491142273},{"id":"https://openalex.org/C118615104","wikidata":"https://www.wikidata.org/wiki/Q121416","display_name":"Discrete mathematics","level":1,"score":0.328000009059906},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.3255999982357025},{"id":"https://openalex.org/C162392398","wikidata":"https://www.wikidata.org/wiki/Q272404","display_name":"Finite set","level":2,"score":0.3239000141620636},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3052000105381012},{"id":"https://openalex.org/C2777299769","wikidata":"https://www.wikidata.org/wiki/Q3707858","display_name":"Type (biology)","level":2,"score":0.2937999963760376},{"id":"https://openalex.org/C206345919","wikidata":"https://www.wikidata.org/wiki/Q20380951","display_name":"Resource (disambiguation)","level":2,"score":0.29260000586509705},{"id":"https://openalex.org/C136119220","wikidata":"https://www.wikidata.org/wiki/Q1000660","display_name":"Algebra over a field","level":2,"score":0.2563999891281128},{"id":"https://openalex.org/C89611455","wikidata":"https://www.wikidata.org/wiki/Q6804646","display_name":"Mechanism (biology)","level":2,"score":0.2522999942302704}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.33835","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.33835","pdf_url":"https://arxiv.org/pdf/2609.33835","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.2609.33835","doi":"https://doi.org/10.48550/arxiv.2609.33835","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.33835","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:2609.33835","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.33835","pdf_url":"https://arxiv.org/pdf/2609.33835","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":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality education","score":0.4156000018119812}],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7214799486.pdf","grobid_xml":"https://content.openalex.org/works/W7214799486.grobid-xml"},"referenced_works_count":11,"referenced_works":["https://openalex.org/W4389911544","https://openalex.org/W4391555764","https://openalex.org/W4407091263","https://openalex.org/W4407123788","https://openalex.org/W4407310244","https://openalex.org/W4407310283","https://openalex.org/W4407570807","https://openalex.org/W4414856916","https://openalex.org/W4415335870","https://openalex.org/W7134271190","https://openalex.org/W7160610743"],"related_works":["https://openalex.org/W6979343220","https://openalex.org/W7134271190","https://openalex.org/W4414856916","https://openalex.org/W4407310244","https://openalex.org/W4415335870","https://openalex.org/W7160610743","https://openalex.org/W7102727821","https://openalex.org/W4403964154","https://openalex.org/W4415797796","https://openalex.org/W4415335946"],"abstract_inverted_index":{"Weak-to-strong":[0],"generalization":[1],"(W2SG)":[2],"occurs":[3,181,193],"when":[4,17,94],"a":[5,9,49,58,130,155,186,209,250],"student":[6,38,59,89,107,161,197],"trained":[7,61],"on":[8,63,67],"teacher's":[10,65,117],"predictions":[11,66],"outperforms":[12,90],"that":[13,36,87,241],"teacher.":[14,44],"We":[15,152],"study":[16,154],"this":[18,124],"happens":[19],"under":[20],"fully":[21],"converged,":[22],"ridgeless":[23],"two-stage":[24,46,251],"learning,":[25],"with":[26,253],"no":[27,30,34,254],"early":[28],"stopping,":[29],"explicit":[31,99,176,202,255],"regularization,":[32],"and":[33,57,80,136,169,189,244],"assumption":[35],"the":[37,43,64,76,81,88,91,106,116,133,142,160,167,196,230],"is":[39,51,60,218,225],"more":[40,164],"expressive":[41],"than":[42,166],"In":[45],"linear":[47],"regression,":[48],"teacher":[50,92,223],"fit":[52],"from":[53],"$n$":[54],"labeled":[55],"examples":[56],"solely":[62],"$m$":[68,95,184,215],"fresh,":[69],"unlabeled":[70],"inputs.":[71],"Although":[72],"both":[73],"stages":[74],"share":[75],"same":[77,82],"hypothesis":[78],"class":[79],"training":[83],"rule,":[84],"we":[85,122],"show":[86,240],"exactly":[93],"lies":[96],"in":[97,126,158,185],"an":[98,201],"intermediate":[100],"range:":[101],"too":[102,111],"few":[103],"pseudo-labels":[104],"leave":[105],"without":[108],"enough":[109],"signal,":[110],"many":[112],"let":[113],"it":[114,192],"inherit":[115],"noise.":[118,236],"Under":[119],"power-law":[120],"covariance,":[121],"derive":[123],"range":[125],"closed":[127],"form":[128],"as":[129],"function":[131],"of":[132,150,214],"spectral":[134],"decay":[135],"noise":[137],"level,":[138],"including":[139],"regimes":[140,205],"where":[141,179,191],"improving":[143],"region":[144],"splits":[145],"into":[146],"two":[147,171],"disjoint":[148],"intervals":[149],"$m$.":[151],"then":[153],"random-feature":[156],"model":[157],"which":[159],"has":[162],"strictly":[163],"features":[165],"teacher,":[168],"identify":[170],"regimes,":[172],"again":[173],"given":[174],"by":[175,208],"thresholds:":[177],"one":[178,190],"improvement":[180],"only":[182,194,233],"for":[183],"bounded":[187],"interval,":[188],"once":[195],"width":[198,246],"$N_S$":[199,217],"exceeds":[200],"threshold.":[203],"Both":[204],"are":[206],"governed":[207],"single":[210],"\"active-bottleneck\"":[211],"principle:":[212],"whichever":[213],"or":[216],"scarcer":[219],"controls":[220],"how":[221],"much":[222],"error":[224],"filtered":[226],"out,":[227],"while":[228],"increasing":[229],"other":[231],"resource":[232],"reduces":[234],"estimation":[235],"Together,":[237],"these":[238],"results":[239],"finite":[242,245],"data":[243],"can":[247],"themselves":[248],"regularize":[249],"learner,":[252],"mechanism":[256],"doing":[257],"so.":[258]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-09-30T00:00:00"}