{"id":"https://openalex.org/W7160658221","doi":"https://doi.org/10.48550/arxiv.2605.05777","title":"Estimating the Black-box LLM Uncertainty with Distribution-Aligned Adversarial Distillation","display_name":"Estimating the Black-box LLM Uncertainty with Distribution-Aligned Adversarial Distillation","publication_year":2026,"publication_date":"2026-05-07","ids":{"openalex":"https://openalex.org/W7160658221","doi":"https://doi.org/10.48550/arxiv.2605.05777"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.05777","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.05777","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.05777","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5135699194","display_name":"Huizi Cui","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cui, Huizi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135656126","display_name":"Huan Ma","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ma, Huan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100710384","display_name":"Qilin Wang","orcid":"https://orcid.org/0000-0001-7968-051X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Qilin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5060896474","display_name":"Yuhang Gao","orcid":"https://orcid.org/0009-0006-1645-2636"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gao, Yuhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135727825","display_name":"Changqing Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Changqing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"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/T10028","display_name":"Topic Modeling","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/T10028","display_name":"Topic Modeling","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/T11636","display_name":"Artificial Intelligence in Healthcare and Education","score":0.11599999666213989,"subfield":{"id":"https://openalex.org/subfields/2718","display_name":"Health Informatics"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}},{"id":"https://openalex.org/T13629","display_name":"Text Readability and Simplification","score":0.06790000200271606,"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/adversarial-system","display_name":"Adversarial system","score":0.8183000087738037},{"id":"https://openalex.org/keywords/bottleneck","display_name":"Bottleneck","score":0.6733999848365784},{"id":"https://openalex.org/keywords/proxy","display_name":"Proxy (statistics)","score":0.659500002861023},{"id":"https://openalex.org/keywords/sampling","display_name":"Sampling (signal processing)","score":0.382099986076355},{"id":"https://openalex.org/keywords/a-priori-and-a-posteriori","display_name":"A priori and a posteriori","score":0.3732999861240387},{"id":"https://openalex.org/keywords/distillation","display_name":"Distillation","score":0.35109999775886536},{"id":"https://openalex.org/keywords/uncertainty-quantification","display_name":"Uncertainty quantification","score":0.3400000035762787}],"concepts":[{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.8183000087738037},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.6733999848365784},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6699000000953674},{"id":"https://openalex.org/C2780148112","wikidata":"https://www.wikidata.org/wiki/Q1432581","display_name":"Proxy (statistics)","level":2,"score":0.659500002861023},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4271000027656555},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.424699991941452},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.382099986076355},{"id":"https://openalex.org/C75553542","wikidata":"https://www.wikidata.org/wiki/Q178161","display_name":"A priori and a posteriori","level":2,"score":0.3732999861240387},{"id":"https://openalex.org/C204030448","wikidata":"https://www.wikidata.org/wiki/Q101017","display_name":"Distillation","level":2,"score":0.35109999775886536},{"id":"https://openalex.org/C32230216","wikidata":"https://www.wikidata.org/wiki/Q7882499","display_name":"Uncertainty quantification","level":2,"score":0.3400000035762787},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3082999885082245},{"id":"https://openalex.org/C20162079","wikidata":"https://www.wikidata.org/wiki/Q1151406","display_name":"Case-based reasoning","level":2,"score":0.30820000171661377},{"id":"https://openalex.org/C167723999","wikidata":"https://www.wikidata.org/wiki/Q3773214","display_name":"Sampling distribution","level":2,"score":0.2888999879360199},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.28850001096725464},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.27070000767707825},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.2639999985694885},{"id":"https://openalex.org/C149441793","wikidata":"https://www.wikidata.org/wiki/Q200726","display_name":"Probability distribution","level":2,"score":0.26089999079704285},{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.26089999079704285},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2515999972820282}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.05777","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.05777","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.05777","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.05777","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"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"},"sustainable_development_goals":[{"display_name":"Reduced inequalities","id":"https://metadata.un.org/sdg/10","score":0.6093387603759766}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"have":[4,138],"progressed":[5],"rapidly":[6],"in":[7,57],"complex":[8],"reasoning":[9,60],"and":[10,51,127,142],"question":[11],"answering,":[12],"yet":[13],"LLM":[14,108,126],"hallucination":[15],"remains":[16],"a":[17,74,79,150],"central":[18],"bottleneck":[19],"that":[20,149,155],"hinders":[21],"practical":[22],"deployment,":[23],"especially":[24],"for":[25,158],"commercial":[26],"black-box":[27,59,94,107,125],"LLMs":[28],"accessible":[29],"only":[30,156],"via":[31],"APIs.":[32],"Existing":[33],"uncertainty":[34,131,168],"quantification":[35],"methods":[36],"typically":[37],"depend":[38],"on":[39,133],"computationally":[40],"expensive":[41],"multiple":[42],"sampling":[43],"or":[44,110],"internal":[45],"parameters,":[46],"which":[47,72],"prevents":[48],"real-time":[49],"estimation":[50],"fails":[52],"to":[53,77,83,103,118],"capture":[54],"information":[55],"implicit":[56],"the":[58,85,89,93,101,106,115,120,124,129,140,161],"process.":[61],"To":[62],"address":[63],"this":[64],"issue,":[65],"we":[66,113],"propose":[67],"Distribution-Aligned":[68],"Adversarial":[69],"Distillation":[70],"(DisAAD),":[71],"introduces":[73],"generation-discrimination":[75],"architecture":[76],"guide":[78],"lightweight":[80],"proxy":[81,116,151],"model":[82,117,152],"learn":[84],"high-quality":[86],"regions":[87],"of":[88,92,123,144,160],"output":[90],"distribution":[91],"LLM,":[95],"thus":[96],"effectively":[97],"endowing":[98],"it":[99],"with":[100],"ability":[102],"know":[104],"whether":[105],"knows":[109],"not.":[111],"Subsequently,":[112],"use":[114],"reproduce":[119],"specific":[121],"responses":[122],"estimate":[128],"corresponding":[130],"based":[132],"evidence":[134],"learning.":[135],"Extensive":[136],"experiments":[137],"verified":[139],"effectiveness":[141],"promise":[143],"our":[145],"proposed":[146],"method,":[147],"indicating":[148],"even":[153],"one":[154],"accounts":[157],"1\\%":[159],"target":[162],"LLM's":[163],"size":[164],"can":[165],"achieve":[166],"reliable":[167],"quantification.":[169]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-09T00:00:00"}