{"id":"https://openalex.org/W4414692581","doi":"https://doi.org/10.1109/tifs.2025.3613882","title":"Amplifying Training Data Exposure Through Fine-Tuning With Pseudo-Labeled Memberships","display_name":"Amplifying Training Data Exposure Through Fine-Tuning With Pseudo-Labeled Memberships","publication_year":2025,"publication_date":"2025-10-01","ids":{"openalex":"https://openalex.org/W4414692581","doi":"https://doi.org/10.1109/tifs.2025.3613882"},"language":"en","primary_location":{"id":"doi:10.1109/tifs.2025.3613882","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tifs.2025.3613882","pdf_url":null,"source":{"id":"https://openalex.org/S61310614","display_name":"IEEE Transactions on Information Forensics and Security","issn_l":"1556-6013","issn":["1556-6013","1556-6021"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Information Forensics and Security","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Myunggyo Oh","orcid":"https://orcid.org/0000-0002-0253-1580"},"institutions":[{"id":"https://openalex.org/I4210095654","display_name":"Korea Pharma (South Korea)","ror":"https://ror.org/00q650071","country_code":"KR","type":"company","lineage":["https://openalex.org/I4210095654"]}],"countries":["KR"],"is_corresponding":true,"raw_author_name":"Myunggyo Oh","raw_affiliation_strings":["KT Research and Development Center, Seoul, South Korea","KT R&D Center, Seoul, Korea"],"raw_orcid":"https://orcid.org/0000-0002-0253-1580","affiliations":[{"raw_affiliation_string":"KT Research and Development Center, Seoul, South Korea","institution_ids":[]},{"raw_affiliation_string":"KT R&D Center, Seoul, Korea","institution_ids":["https://openalex.org/I4210095654"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5108390820","display_name":"Hong Eun Ahn","orcid":null},"institutions":[{"id":"https://openalex.org/I193775966","display_name":"Yonsei University","ror":"https://ror.org/01wjejq96","country_code":"KR","type":"education","lineage":["https://openalex.org/I193775966"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Hong Eun Ahn","raw_affiliation_strings":["AI Security and Information Security Laboratory, Yonsei University, Seoul, South Korea","Yonsei University, Seoul, Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"AI Security and Information Security Laboratory, Yonsei University, Seoul, South Korea","institution_ids":["https://openalex.org/I193775966"]},{"raw_affiliation_string":"Yonsei University, Seoul, Korea","institution_ids":["https://openalex.org/I193775966"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5024152438","display_name":"Leo Hyun Park","orcid":null},"institutions":[{"id":"https://openalex.org/I193775966","display_name":"Yonsei University","ror":"https://ror.org/01wjejq96","country_code":"KR","type":"education","lineage":["https://openalex.org/I193775966"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Leo Hyun Park","raw_affiliation_strings":["AI Security and Information Security Laboratory, Yonsei University, Seoul, South Korea","Yonsei University, Seoul, Korea"],"raw_orcid":"https://orcid.org/0000-0002-3100-2258","affiliations":[{"raw_affiliation_string":"AI Security and Information Security Laboratory, Yonsei University, Seoul, South Korea","institution_ids":["https://openalex.org/I193775966"]},{"raw_affiliation_string":"Yonsei University, Seoul, Korea","institution_ids":["https://openalex.org/I193775966"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5090551692","display_name":"Taekyoung Kwon","orcid":"https://orcid.org/0000-0002-5513-0836"},"institutions":[{"id":"https://openalex.org/I193775966","display_name":"Yonsei University","ror":"https://ror.org/01wjejq96","country_code":"KR","type":"education","lineage":["https://openalex.org/I193775966"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Taekyoung Kwon","raw_affiliation_strings":["AI Security and Information Security Laboratory, Yonsei University, Seoul, South Korea","Yonsei University, Seoul, Korea"],"raw_orcid":"https://orcid.org/0000-0002-5513-0836","affiliations":[{"raw_affiliation_string":"AI Security and Information Security Laboratory, Yonsei University, Seoul, South Korea","institution_ids":["https://openalex.org/I193775966"]},{"raw_affiliation_string":"Yonsei University, Seoul, Korea","institution_ids":["https://openalex.org/I193775966"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I4210095654"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.13565975,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"21","issue":null,"first_page":"1363","last_page":"1378"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11902","display_name":"Intelligent Tutoring Systems and Adaptive Learning","score":0.3806000053882599,"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/T11902","display_name":"Intelligent Tutoring Systems and Adaptive Learning","score":0.3806000053882599,"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/training-set","display_name":"Training set","score":0.6643999814987183},{"id":"https://openalex.org/keywords/selection","display_name":"Selection (genetic algorithm)","score":0.6607000231742859},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.6309000253677368},{"id":"https://openalex.org/keywords/data-extraction","display_name":"Data extraction","score":0.5177000164985657},{"id":"https://openalex.org/keywords/labeled-data","display_name":"Labeled data","score":0.4359000027179718},{"id":"https://openalex.org/keywords/selection-bias","display_name":"Selection bias","score":0.3928999900817871}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8158000111579895},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.6643999814987183},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.6607000231742859},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.6309000253677368},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.54339998960495},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.531000018119812},{"id":"https://openalex.org/C2777466982","wikidata":"https://www.wikidata.org/wiki/Q5227287","display_name":"Data extraction","level":3,"score":0.5177000164985657},{"id":"https://openalex.org/C2776145971","wikidata":"https://www.wikidata.org/wiki/Q30673951","display_name":"Labeled data","level":2,"score":0.4359000027179718},{"id":"https://openalex.org/C40423286","wikidata":"https://www.wikidata.org/wiki/Q284172","display_name":"Selection bias","level":2,"score":0.3928999900817871},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.3837999999523163},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.38370001316070557},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.35089999437332153},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.32499998807907104},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3003000020980835},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.29679998755455017},{"id":"https://openalex.org/C90559484","wikidata":"https://www.wikidata.org/wiki/Q778379","display_name":"Expression (computer science)","level":2,"score":0.2685999870300293},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.2678000032901764}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tifs.2025.3613882","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tifs.2025.3613882","pdf_url":null,"source":{"id":"https://openalex.org/S61310614","display_name":"IEEE Transactions on Information Forensics and Security","issn_l":"1556-6013","issn":["1556-6013","1556-6021"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Information Forensics and Security","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G3605687782","display_name":null,"funder_award_id":"RS-2024-00439762","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"},{"id":"https://openalex.org/G4291211624","display_name":null,"funder_award_id":"RS-2025-02304983","funder_id":"https://openalex.org/F4320334869","funder_display_name":"Korean National Police Agency"}],"funders":[{"id":"https://openalex.org/F4320334869","display_name":"Korean National Police Agency","ror":"https://ror.org/05x57gp50"},{"id":"https://openalex.org/F4320335199","display_name":"Korea Institute of Energy Technology Evaluation and Planning","ror":"https://ror.org/02zq38y32"},{"id":"https://openalex.org/F4320335489","display_name":"Institute for Information and Communications Technology Promotion","ror":"https://ror.org/01g0hqq23"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":35,"referenced_works":["https://openalex.org/W1566289585","https://openalex.org/W1682403713","https://openalex.org/W2158874082","https://openalex.org/W2473418344","https://openalex.org/W2535690855","https://openalex.org/W2795435272","https://openalex.org/W2963456134","https://openalex.org/W2965373594","https://openalex.org/W2970482702","https://openalex.org/W2979826702","https://openalex.org/W2987861506","https://openalex.org/W3002330681","https://openalex.org/W3034287667","https://openalex.org/W3046357466","https://openalex.org/W3081168214","https://openalex.org/W3081488247","https://openalex.org/W3088059392","https://openalex.org/W3091002423","https://openalex.org/W3165327186","https://openalex.org/W3169483174","https://openalex.org/W3171397222","https://openalex.org/W3177765786","https://openalex.org/W3188505388","https://openalex.org/W4230167402","https://openalex.org/W4241006348","https://openalex.org/W4288057780","https://openalex.org/W4308410483","https://openalex.org/W4384345715","https://openalex.org/W4385565597","https://openalex.org/W4385573004","https://openalex.org/W4385573947","https://openalex.org/W4385894687","https://openalex.org/W4388312423","https://openalex.org/W4405181291","https://openalex.org/W4415795414"],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"are":[4,85],"vulnerable":[5],"to":[6,12,29,51,66,80,147,179],"training":[7,36,183],"data":[8,13,104,184],"extraction":[9],"attacks":[10],"due":[11],"memorization.":[14],"This":[15],"paper":[16],"introduces":[17],"a":[18,56,168,177],"novel":[19],"attack":[20],"scenario":[21],"wherein":[22],"an":[23],"attacker":[24,78],"adversarially":[25],"fine-tunes":[26],"pre-trained":[27],"LLMs":[28,171],"amplify":[30],"the":[31,34,63,71,77,89,96,100,115,131,138,156],"exposure":[32],"of":[33,70,95,102,117,153],"original":[35],"data.":[37,91],"Unlike":[38],"prior":[39],"TDE":[40],"methods":[41],"that":[42,84],"mainly":[43],"rely":[44],"on":[45,160],"post-hoc":[46],"querying":[47],"or":[48],"prompt":[49],"selection":[50],"elicit":[52],"memorized":[53],"content":[54],"from":[55,130,143,155],"fixed":[57],"model,":[58],"our":[59],"strategy":[60],"directly":[61],"alters":[62],"model\u2019s":[64],"parameters":[65,175],"intensify":[67],"its":[68],"retention":[69],"pre-training":[72,90,103,157],"dataset.":[73],"To":[74,110],"achieve":[75],"this,":[76,112],"needs":[79],"collect":[81],"generated":[82,106,121],"texts":[83,107],"closely":[86],"aligned":[87],"with":[88,150,172],"However,":[92],"without":[93],"knowledge":[94],"actual":[97],"dataset,":[98],"quantifying":[99],"amount":[101],"within":[105],"is":[108],"challenging.":[109],"address":[111],"we":[113],"propose":[114],"use":[116],"pseudo-labels":[118],"for":[119],"these":[120,161],"texts,":[122],"leveraging":[123],"membership":[124,162],"approximations":[125],"indicated":[126],"by":[127],"machine-generated":[128],"probabilities":[129],"target":[132],"LLMusing":[133],"DetectGPT.":[134],"We":[135,186],"subsequently":[136],"fine-tune":[137],"LLM":[139],"via":[140],"reinforcement":[141],"learning":[142],"human":[144],"feedback":[145],"(RLHF)":[146],"favor":[148],"generations":[149],"higher":[151],"likelihoods":[152],"originating":[154],"data,":[158],"based":[159],"probabilities.":[163],"Our":[164],"empirical":[165],"findings":[166],"indicate":[167],"remarkable":[169],"outcome:":[170],"over":[173],"1B":[174],"exhibit":[176],"four":[178],"eight-fold":[180],"increase":[181],"in":[182],"exposure.":[185],"discuss":[187],"potential":[188],"mitigations":[189],"and":[190],"suggest":[191],"future":[192],"research":[193],"directions.":[194]},"counts_by_year":[],"updated_date":"2026-02-07T06:11:34.122080","created_date":"2025-10-10T00:00:00"}