{"id":"https://openalex.org/W4387929697","doi":"https://doi.org/10.48550/arxiv.2310.15007","title":"Did the Neurons Read your Book? Document-level Membership Inference for Large Language Models","display_name":"Did the Neurons Read your Book? Document-level Membership Inference for Large Language Models","publication_year":2023,"publication_date":"2023-10-23","ids":{"openalex":"https://openalex.org/W4387929697","doi":"https://doi.org/10.48550/arxiv.2310.15007"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2310.15007","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2310.15007","pdf_url":"https://arxiv.org/pdf/2310.15007","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":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2310.15007","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5068554561","display_name":"Matthieu Meeus","orcid":"https://orcid.org/0009-0008-7353-4042","observed_orcids":["https://orcid.org/0009-0008-7353-4042"]},"institutions":[{"id":"https://openalex.org/I47508984","display_name":"Imperial College London","ror":"https://ror.org/041kmwe10","country_code":"GB","type":"education","lineage":["https://openalex.org/I47508984"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Meeus, Matthieu","raw_affiliation_strings":["Imperial College London"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Imperial College London","institution_ids":["https://openalex.org/I47508984"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5001915349","display_name":"Shubham Jain","orcid":"https://orcid.org/0000-0002-7978-9504","observed_orcids":["https://orcid.org/0000-0002-7978-9504"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jain, Shubham","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5062209641","display_name":"Marek Rei","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I47508984","display_name":"Imperial College London","ror":"https://ror.org/041kmwe10","country_code":"GB","type":"education","lineage":["https://openalex.org/I47508984"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Rei, Marek","raw_affiliation_strings":["Imperial College London"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Imperial College London","institution_ids":["https://openalex.org/I47508984"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5078253058","display_name":"Yves-Alexandre de Montjoye","orcid":"https://orcid.org/0000-0002-2559-5616","observed_orcids":["https://orcid.org/0000-0002-2559-5616"]},"institutions":[{"id":"https://openalex.org/I47508984","display_name":"Imperial College London","ror":"https://ror.org/041kmwe10","country_code":"GB","type":"education","lineage":["https://openalex.org/I47508984"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"de Montjoye, Yves-Alexandre","raw_affiliation_strings":["Imperial College London"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Imperial College London","institution_ids":["https://openalex.org/I47508984"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":5,"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/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.6342999935150146,"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/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.6342999935150146,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.2782999873161316,"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":0.04100000113248825,"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/membership-inference","display_name":"membership inference","score":0.7200000286102295},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.9100000262260437},{"id":"https://openalex.org/keywords/training-data-privacy","display_name":"training data privacy","score":0.7360000014305115},{"id":"https://openalex.org/keywords/black-box-attacks","display_name":"black-box attacks","score":0.699999988079071},{"id":"https://openalex.org/keywords/model-size-effects","display_name":"model size effects","score":0.45100000500679016},{"id":"https://openalex.org/keywords/mitigation-measures","display_name":"mitigation measures","score":0.6549999713897705}],"concepts":[{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.7305316925048828},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.577905535697937},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.544032633304596},{"id":"https://openalex.org/C2780233690","wikidata":"https://www.wikidata.org/wiki/Q535347","display_name":"Transparency (behavior)","level":2,"score":0.47043582797050476},{"id":"https://openalex.org/C2777530160","wikidata":"https://www.wikidata.org/wiki/Q41796","display_name":"Sentence","level":2,"score":0.4587893486022949},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.42232826352119446},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3995491862297058},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.12367069721221924},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.11871343851089478},{"id":"https://openalex.org/C201995342","wikidata":"https://www.wikidata.org/wiki/Q682496","display_name":"Systems engineering","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2310.15007","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2310.15007","pdf_url":"https://arxiv.org/pdf/2310.15007","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":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2310.15007","doi":"https://doi.org/10.48550/arxiv.2310.15007","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2310.15007","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2310.15007","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2310.15007","pdf_url":"https://arxiv.org/pdf/2310.15007","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":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4387929697.pdf","grobid_xml":"https://content.openalex.org/works/W4387929697.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4411120109","https://openalex.org/W7165643232","https://openalex.org/W7220383128","https://openalex.org/W7155164840","https://openalex.org/W7171555260","https://openalex.org/W4410609172","https://openalex.org/W7168040614","https://openalex.org/W7125800646","https://openalex.org/W4414763629","https://openalex.org/W4404783485"],"abstract_inverted_index":{"With":[0],"large":[1],"language":[2],"models":[3,194],"(LLMs)":[4],"poised":[5,264],"to":[6,16,40,64,133,152,172,199,205,212,225,235,265],"become":[7,61],"embedded":[8],"in":[9,180],"our":[10,150,170,213,247,267],"daily":[11],"lives,":[12],"questions":[13,26,41,53],"are":[14,232],"starting":[15],"be":[17,196,206,255],"raised":[18],"about":[19],"the":[20,57,74,86,103,122,174,181,185,223,240,260],"data":[21,39,117],"they":[22],"learned":[23],"from.":[24],"These":[25],"range":[27],"from":[28,36],"potential":[29],"bias":[30],"or":[31,95],"misinformation":[32],"LLMs":[33,60,112],"could":[34],"retain":[35],"their":[37,68],"training":[38,69,94,120],"of":[42,47,56,76,107,159,262],"copyright":[43],"and":[44,105,121,137,145,163,202,221],"fair":[45],"use":[46],"human-generated":[48],"text.":[49],"However,":[50],"while":[51],"these":[52],"emerge,":[54],"developers":[55],"recent":[58],"state-of-the-art":[59],"increasingly":[62],"reluctant":[63],"disclose":[65],"details":[66],"on":[67,140],"corpus.":[70],"We":[71,126,148,167,189],"here":[72],"introduce":[73],"task":[75],"document-level":[77,108,135,186,200,252],"membership":[78,109,136,176,187,253],"inference":[79,110,177,201],"for":[80,102,111,119,161,165,184,257],"real-world":[81],"LLMs,":[82,258],"i.e.":[83],"inferring":[84],"whether":[85,192],"LLM":[87],"has":[88],"seen":[89],"a":[90,100,129],"given":[91],"document":[92],"during":[93],"not.":[96],"First,":[97],"we":[98,216],"propose":[99,128],"procedure":[101],"development":[104],"evaluation":[106],"by":[113],"leveraging":[114],"commonly":[115],"used":[116,179],"sources":[118],"model":[123,241],"release":[124],"date.":[125],"then":[127,168],"practical,":[130],"black-box":[131],"method":[132],"predict":[134],"instantiate":[138],"it":[139],"OpenLLaMA-7B":[141,211],"with":[142],"both":[143],"books":[144,162],"academic":[146],"papers.":[147,166],"show":[149,169,203,249],"methodology":[151],"perform":[153],"very":[154],"well,":[155],"reaching":[156],"an":[157],"AUC":[158,224],"0.856":[160],"0.678":[164],"approach":[171],"outperform":[173],"sentence-level":[175],"attacks":[178],"privacy":[182],"literature":[183],"task.":[188],"further":[190],"evaluate":[191],"smaller":[193],"might":[195],"less":[197],"sensitive":[198,209],"OpenLLaMA-3B":[204],"approximately":[207],"as":[208,210],"approach.":[214],"Finally,":[215],"consider":[217],"two":[218],"mitigation":[219],"strategies":[220],"find":[222],"slowly":[226],"decrease":[227],"when":[228,239],"only":[229],"partial":[230],"documents":[231],"considered":[233],"but":[234],"remain":[236],"fairly":[237],"high":[238],"precision":[242],"is":[243],"reduced.":[244],"Taken":[245],"together,":[246],"results":[248],"that":[250],"accurate":[251],"can":[254],"inferred":[256],"increasing":[259],"transparency":[261],"technology":[263],"change":[266],"lives.":[268]},"counts_by_year":[{"year":2025,"cited_by_count":5}],"updated_date":"2026-10-16T09:59:22.235173","created_date":"2023-10-25T00:00:00"}