{"id":"https://openalex.org/W7117707685","doi":"https://doi.org/10.48550/arxiv.2512.22886","title":"Theory and Algorithms for Learning with Multi-Class Abstention and Multi-Expert Deferral","display_name":"Theory and Algorithms for Learning with Multi-Class Abstention and Multi-Expert Deferral","publication_year":2025,"publication_date":"2025-12-28","ids":{"openalex":"https://openalex.org/W7117707685","doi":"https://doi.org/10.48550/arxiv.2512.22886"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2512.22886","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2512.22886","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,"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2512.22886","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5014674732","display_name":"Anqi Mao","orcid":"https://orcid.org/0000-0003-3062-5311"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Mao, Anqi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5014674732"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"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":0.3009999990463257,"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":0.3009999990463257,"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":0.07760000228881836,"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/T10028","display_name":"Topic Modeling","score":0.07249999791383743,"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/deferral","display_name":"Deferral","score":0.760699987411499},{"id":"https://openalex.org/keywords/consistency","display_name":"Consistency (knowledge bases)","score":0.6025999784469604},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5415999889373779},{"id":"https://openalex.org/keywords/regression","display_name":"Regression","score":0.445499986410141},{"id":"https://openalex.org/keywords/heuristics","display_name":"Heuristics","score":0.41370001435279846},{"id":"https://openalex.org/keywords/face","display_name":"Face (sociological concept)","score":0.3059000074863434}],"concepts":[{"id":"https://openalex.org/C2778281098","wikidata":"https://www.wikidata.org/wiki/Q885383","display_name":"Deferral","level":2,"score":0.760699987411499},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7369999885559082},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.6025999784469604},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5415999889373779},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5145000219345093},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.46380001306533813},{"id":"https://openalex.org/C83546350","wikidata":"https://www.wikidata.org/wiki/Q1139051","display_name":"Regression","level":2,"score":0.445499986410141},{"id":"https://openalex.org/C127705205","wikidata":"https://www.wikidata.org/wiki/Q5748245","display_name":"Heuristics","level":2,"score":0.41370001435279846},{"id":"https://openalex.org/C2779304628","wikidata":"https://www.wikidata.org/wiki/Q3503480","display_name":"Face (sociological concept)","level":2,"score":0.3059000074863434},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.29330000281333923},{"id":"https://openalex.org/C158600405","wikidata":"https://www.wikidata.org/wiki/Q5054566","display_name":"Causal inference","level":2,"score":0.28679999709129333},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.2815999984741211},{"id":"https://openalex.org/C2777027219","wikidata":"https://www.wikidata.org/wiki/Q1284190","display_name":"Constant (computer programming)","level":2,"score":0.2793999910354614},{"id":"https://openalex.org/C152877465","wikidata":"https://www.wikidata.org/wiki/Q208042","display_name":"Regression analysis","level":2,"score":0.2791000008583069},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.26460000872612}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2512.22886","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2512.22886","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,"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":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2512.22886","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2512.22886","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,"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"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2,39],"(LLMs)":[3],"have":[4],"achieved":[5],"remarkable":[6],"performance":[7,130],"but":[8],"face":[9],"critical":[10],"challenges:":[11],"hallucinations":[12],"and":[13,60,86,99,116,125,150,153,204,211,230],"high":[14],"inference":[15],"costs.":[16],"Leveraging":[17],"multiple":[18,202],"experts":[19,29,203],"offers":[20],"a":[21,54,186],"solution:":[22],"deferring":[23],"uncertain":[24],"inputs":[25],"to":[26,36,179,193],"more":[27],"capable":[28],"improves":[30],"reliability,":[31],"while":[32],"routing":[33],"simpler":[34],"queries":[35],"smaller,":[37],"distilled":[38],"enhances":[40],"efficiency.":[41],"This":[42,51],"motivates":[43],"the":[44,61,128,162,232,236],"problem":[45,59,63],"of":[46,57,64,81,96,131,235],"learning":[47,65,75],"with":[48,66,76,120,191,220,227],"multiple-expert":[49],"deferral.":[50],"thesis":[52],"presents":[53],"comprehensive":[55],"study":[56],"this":[58],"related":[62],"abstention,":[67],"supported":[68],"by":[69],"strong":[70,101,158],"consistency":[71,105],"guarantees.":[72],"First,":[73],"for":[74,147,174,189],"abstention":[77],"(a":[78],"special":[79],"case":[80],"deferral),":[82],"we":[83,135,165,184],"analyze":[84,113],"score-based":[85],"predictor-rejector":[87],"formulations":[88],"in":[89,140],"multi-class":[90],"classification.":[91,141],"We":[92,112,142,222],"introduce":[93,185],"new":[94,144,181,224],"families":[95],"surrogate":[97,145,169,225],"losses":[98,146,170,226],"prove":[100,154],"non-asymptotic,":[102],"hypothesis":[103],"set-specific":[104],"guarantees,":[106],"resolving":[107],"two":[108],"existing":[109],"open":[110],"questions.":[111],"both":[114,148,209],"single-stage":[115,149,210],"practical":[117],"two-stage":[118,151,163,212],"settings,":[119],"experiments":[121],"on":[122,218],"CIFAR-10,":[123],"CIFAR-100,":[124],"SVHN":[126],"demonstrating":[127],"superior":[129],"our":[132,168],"algorithms.":[133,182,238],"Second,":[134],"address":[136,194],"general":[137],"multi-expert":[138],"deferral":[139,192],"design":[143],"scenarios":[152],"they":[155],"benefit":[156],"from":[157],"$H$-consistency":[159,229],"bounds.":[160],"For":[161],"scenario,":[164],"show":[166],"that":[167],"are":[171],"realizable":[172],"$H$-consistent":[173],"constant":[175],"cost":[176,206],"functions,":[177],"leading":[178],"effective":[180],"Finally,":[183],"novel":[187],"framework":[188,200],"regression":[190,219],"continuous":[195],"label":[196],"spaces.":[197],"Our":[198],"versatile":[199],"accommodates":[201],"various":[205],"structures,":[207],"supporting":[208],"methods.":[213],"It":[214],"subsumes":[215],"recent":[216],"work":[217],"abstention.":[221],"propose":[223],"proven":[228],"demonstrate":[231],"empirical":[233],"effectiveness":[234],"resulting":[237]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2025-12-31T00:00:00"}