{"id":"https://openalex.org/W7160966664","doi":"https://doi.org/10.48550/arxiv.2605.10407","title":"Identified-Set Geometry of Distributional Model Extraction under Top-$K$ Censored API Access","display_name":"Identified-Set Geometry of Distributional Model Extraction under Top-$K$ Censored API Access","publication_year":2026,"publication_date":"2026-05-11","ids":{"openalex":"https://openalex.org/W7160966664","doi":"https://doi.org/10.48550/arxiv.2605.10407"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.10407","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10407","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.2605.10407","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5135986278","display_name":"Wenhua Nie","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Nie, Wenhua","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135917007","display_name":"ZiCheng Zhu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhu, ZiCheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5006906912","display_name":"Jianan Wu","orcid":"https://orcid.org/0000-0002-6513-4769"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Jianan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135989815","display_name":"Binhan Luo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Luo, Binhan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135947978","display_name":"Haoran Zheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Haoran","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5073451247","display_name":"Jyh\u2010Shing Roger Jang","orcid":"https://orcid.org/0000-0002-7319-9095"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jang, Jyh-Shing Roger","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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.32010000944137573,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.32010000944137573,"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/T11424","display_name":"Security and Verification in Computing","score":0.25760000944137573,"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/T12479","display_name":"Web Application Security Vulnerabilities","score":0.07729999721050262,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/censoring","display_name":"Censoring (clinical trials)","score":0.8009999990463257},{"id":"https://openalex.org/keywords/partition","display_name":"Partition (number theory)","score":0.5968000292778015},{"id":"https://openalex.org/keywords/logit","display_name":"Logit","score":0.5752999782562256},{"id":"https://openalex.org/keywords/upper-and-lower-bounds","display_name":"Upper and lower bounds","score":0.5270000100135803},{"id":"https://openalex.org/keywords/extension","display_name":"Extension (predicate logic)","score":0.46560001373291016},{"id":"https://openalex.org/keywords/matching","display_name":"Matching (statistics)","score":0.4603999853134155},{"id":"https://openalex.org/keywords/mixed-logit","display_name":"Mixed logit","score":0.40149998664855957},{"id":"https://openalex.org/keywords/distribution","display_name":"Distribution (mathematics)","score":0.3718000054359436}],"concepts":[{"id":"https://openalex.org/C137668524","wikidata":"https://www.wikidata.org/wiki/Q189813","display_name":"Censoring (clinical trials)","level":2,"score":0.8009999990463257},{"id":"https://openalex.org/C42812","wikidata":"https://www.wikidata.org/wiki/Q1082910","display_name":"Partition (number theory)","level":2,"score":0.5968000292778015},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.590499997138977},{"id":"https://openalex.org/C140331021","wikidata":"https://www.wikidata.org/wiki/Q1868104","display_name":"Logit","level":2,"score":0.5752999782562256},{"id":"https://openalex.org/C77553402","wikidata":"https://www.wikidata.org/wiki/Q13222579","display_name":"Upper and lower bounds","level":2,"score":0.5270000100135803},{"id":"https://openalex.org/C2778029271","wikidata":"https://www.wikidata.org/wiki/Q5421931","display_name":"Extension (predicate logic)","level":2,"score":0.46560001373291016},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.4603999853134155},{"id":"https://openalex.org/C95057490","wikidata":"https://www.wikidata.org/wiki/Q6883984","display_name":"Mixed logit","level":3,"score":0.40149998664855957},{"id":"https://openalex.org/C110121322","wikidata":"https://www.wikidata.org/wiki/Q865811","display_name":"Distribution (mathematics)","level":2,"score":0.3718000054359436},{"id":"https://openalex.org/C2776175482","wikidata":"https://www.wikidata.org/wiki/Q1195816","display_name":"Transfer (computing)","level":2,"score":0.3675000071525574},{"id":"https://openalex.org/C204030448","wikidata":"https://www.wikidata.org/wiki/Q101017","display_name":"Distillation","level":2,"score":0.366100013256073},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.3587999939918518},{"id":"https://openalex.org/C4725764","wikidata":"https://www.wikidata.org/wiki/Q844704","display_name":"Extraction (chemistry)","level":2,"score":0.3582000136375427},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.35510000586509705},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.33709999918937683},{"id":"https://openalex.org/C49781872","wikidata":"https://www.wikidata.org/wiki/Q1045555","display_name":"Maximum likelihood","level":2,"score":0.32760000228881836},{"id":"https://openalex.org/C149441793","wikidata":"https://www.wikidata.org/wiki/Q200726","display_name":"Probability distribution","level":2,"score":0.3264999985694885},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.3231000006198883},{"id":"https://openalex.org/C2776459999","wikidata":"https://www.wikidata.org/wiki/Q2119376","display_name":"Fidelity","level":2,"score":0.32260000705718994},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.30820000171661377},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.29580000042915344},{"id":"https://openalex.org/C113364801","wikidata":"https://www.wikidata.org/wiki/Q26674","display_name":"High fidelity","level":2,"score":0.2754000127315521},{"id":"https://openalex.org/C151956035","wikidata":"https://www.wikidata.org/wiki/Q1132755","display_name":"Logistic regression","level":2,"score":0.2551000118255615}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.10407","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10407","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.2605.10407","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10407","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":[{"score":0.4925192892551422,"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Modern":[0],"LLM":[1],"APIs":[2],"often":[3],"reveal":[4,78],"only":[5],"top-$K$":[6,84],"logit":[7,125],"scores":[8],"and":[9,59,99],"censor":[10],"the":[11,16,28,45],"remaining":[12],"vocabulary.":[13],"We":[14],"study":[15],"per-position":[17,108],"distribution-recovery":[18],"limits":[19,107],"of":[20,88],"this":[21],"access":[22],"model.":[23],"For":[24,49],"censoring":[25,105],"threshold":[26],"$\u03c4$,":[27],"compatible":[29],"teacher":[30,77],"distributions":[31],"form":[32],"an":[33,60,67],"identified":[34],"set":[35],"whose":[36],"total-variation":[37],"diameter":[38],"is":[39,44],"exactly":[40],"$U_K=(V-K)\\exp(\u03c4)/(Z_A+(V-K)\\exp(\u03c4))$,":[41],"where":[42],"$Z_A$":[43],"observed":[46],"partition":[47],"function.":[48],"KL":[50,97],"recovery,":[51],"we":[52],"give":[53],"a":[54,74,79],"computable":[55],"binary-endpoint":[56],"lower":[57],"bound":[58],"asymptotically":[61],"matching":[62],"small-ambiguity":[63],"upper":[64],"bound,":[65],"with":[66],"extension":[68],"to":[69],"reference-aware":[70],"attackers.":[71],"Experiments":[72],"on":[73],"Qwen3":[75],"math-reasoning":[76],"layered":[80],"extraction":[81,101],"hierarchy:":[82],"on-task":[83],"distillation":[85,92],"recovers":[86,93,102],"12%":[87],"private":[89],"capability,":[90],"full-logit":[91],"56%":[94],"despite":[95],"99%":[96],"closure,":[98],"generation-based":[100],"96%.":[103],"Top-$K$":[104],"therefore":[106],"distribution":[109],"recovery":[110],"but":[111],"does":[112],"not":[113],"by":[114],"itself":[115],"prevent":[116],"capability":[117],"extraction,":[118],"separating":[119],"fidelity":[120],"from":[121],"transfer":[122],"in":[123],"prompt-only":[124],"distillation.":[126]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-05-13T00:00:00"}