{"id":"https://openalex.org/W7131670425","doi":"https://doi.org/10.48550/arxiv.2602.21218","title":"EPSVec: Efficient and Private Synthetic Data Generation via Dataset Vectors","display_name":"EPSVec: Efficient and Private Synthetic Data Generation via Dataset Vectors","publication_year":2026,"publication_date":"2026-01-31","ids":{"openalex":"https://openalex.org/W7131670425","doi":"https://doi.org/10.48550/arxiv.2602.21218"},"language":null,"primary_location":{"id":"pmh:doi:10.48550/arxiv.2602.21218","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"type":"article","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":null,"any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5116235417","display_name":"Amin Banayeeanzade","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Banayeeanzade, Amin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126936513","display_name":"Qingchuan Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Qingchuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113113923","display_name":"Deqing Fu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fu, Deqing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126861430","display_name":"Spencer Hong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hong, Spencer","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126895537","display_name":"Erin Babinsky","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Babinsky, Erin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5066674792","display_name":"Alfy Samuel","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Samuel, Alfy","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126947660","display_name":"Anoop Kumar","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kumar, Anoop","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5126911334","display_name":"Robin Jia","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jia, Robin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5126917123","display_name":"Sai Praneeth Karimireddy","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Karimireddy, Sai Praneeth","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":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.24012473,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"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":0.29269999265670776,"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":0.29269999265670776,"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":0.11540000140666962,"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/T11704","display_name":"Mobile Crowdsensing and Crowdsourcing","score":0.05979999899864197,"subfield":{"id":"https://openalex.org/subfields/1706","display_name":"Computer Science Applications"},"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/usable","display_name":"USable","score":0.746399998664856},{"id":"https://openalex.org/keywords/synthetic-data","display_name":"Synthetic data","score":0.6481999754905701},{"id":"https://openalex.org/keywords/downstream","display_name":"Downstream (manufacturing)","score":0.5770000219345093},{"id":"https://openalex.org/keywords/fidelity","display_name":"Fidelity","score":0.5756999850273132},{"id":"https://openalex.org/keywords/space","display_name":"Space (punctuation)","score":0.38690000772476196},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.37400001287460327},{"id":"https://openalex.org/keywords/high-fidelity","display_name":"High fidelity","score":0.3465999960899353},{"id":"https://openalex.org/keywords/data-modeling","display_name":"Data modeling","score":0.33820000290870667}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7926999926567078},{"id":"https://openalex.org/C2780615836","wikidata":"https://www.wikidata.org/wiki/Q2471869","display_name":"USable","level":2,"score":0.746399998664856},{"id":"https://openalex.org/C160920958","wikidata":"https://www.wikidata.org/wiki/Q7662746","display_name":"Synthetic data","level":2,"score":0.6481999754905701},{"id":"https://openalex.org/C2776207758","wikidata":"https://www.wikidata.org/wiki/Q5303302","display_name":"Downstream (manufacturing)","level":2,"score":0.5770000219345093},{"id":"https://openalex.org/C2776459999","wikidata":"https://www.wikidata.org/wiki/Q2119376","display_name":"Fidelity","level":2,"score":0.5756999850273132},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.4092000126838684},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4032000005245209},{"id":"https://openalex.org/C2778572836","wikidata":"https://www.wikidata.org/wiki/Q380933","display_name":"Space (punctuation)","level":2,"score":0.38690000772476196},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.37400001287460327},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.37310001254081726},{"id":"https://openalex.org/C113364801","wikidata":"https://www.wikidata.org/wiki/Q26674","display_name":"High fidelity","level":2,"score":0.3465999960899353},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.33820000290870667},{"id":"https://openalex.org/C99221444","wikidata":"https://www.wikidata.org/wiki/Q1532069","display_name":"Private information retrieval","level":2,"score":0.31690001487731934},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.29980000853538513},{"id":"https://openalex.org/C8505890","wikidata":"https://www.wikidata.org/wiki/Q605095","display_name":"Budget constraint","level":2,"score":0.2978000044822693},{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.2921000123023987},{"id":"https://openalex.org/C2776187449","wikidata":"https://www.wikidata.org/wiki/Q1513879","display_name":"Natural language generation","level":3,"score":0.28870001435279846},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.27459999918937683},{"id":"https://openalex.org/C167927819","wikidata":"https://www.wikidata.org/wiki/Q1930567","display_name":"Shuffling","level":2,"score":0.2720000147819519},{"id":"https://openalex.org/C35639132","wikidata":"https://www.wikidata.org/wiki/Q7452468","display_name":"Sequence labeling","level":3,"score":0.26759999990463257},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.26190000772476196},{"id":"https://openalex.org/C137822555","wikidata":"https://www.wikidata.org/wiki/Q2587068","display_name":"Information sensitivity","level":2,"score":0.2565999925136566},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.25220000743865967}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:doi:10.48550/arxiv.2602.21218","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},{"id":"doi:10.48550/arxiv.2602.21218","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.21218","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":"pmh:doi:10.48550/arxiv.2602.21218","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Article"},"sustainable_development_goals":[{"score":0.44357460737228394,"display_name":"Industry, innovation and infrastructure","id":"https://metadata.un.org/sdg/9"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"High-quality":[0],"data":[1,20,92],"is":[2],"essential":[3],"for":[4,25,38],"modern":[5],"machine":[6],"learning,":[7],"yet":[8],"many":[9,118],"valuable":[10],"corpora":[11,60],"are":[12,47,51],"sensitive":[13],"and":[14,28,55,93,98,104,125,143,151,164],"cannot":[15],"be":[16],"freely":[17],"shared.":[18],"Synthetic":[19],"offers":[21],"a":[22,71],"practical":[23],"substitute":[24],"downstream":[26,165],"development,":[27],"large":[29,58],"language":[30],"models":[31,142],"(LLMs)":[32],"have":[33],"emerged":[34],"as":[35],"powerful":[36],"engines":[37],"generating":[39],"it.":[40],"However,":[41],"existing":[42,159],"private":[43,59,91],"text":[44],"generation":[45,78,149],"methods":[46],"severely":[48],"inefficient:":[49],"they":[50],"data-intensive,":[52],"computationally":[53],"slow,":[54],"often":[56],"require":[57],"or":[61],"batch":[62],"sizes":[63],"to":[64,147],"achieve":[65],"usable":[66],"quality.":[67],"We":[68],"introduce":[69],"EPSVec,":[70],"differentially-private":[72],"lightweight":[73],"alternative":[74],"that":[75,85,156],"steers":[76],"LLM":[77],"using":[79],"*dataset":[80],"vectors*--directions":[81],"in":[82,130,161,168],"activation":[83],"space":[84],"capture":[86],"the":[87,111],"distributional":[88,162],"gap":[89],"between":[90],"public":[94],"priors.":[95],"EPSVec":[96,157],"extracts":[97],"sanitizes":[99],"steering":[100],"vectors":[101],"just":[102],"once":[103],"then":[105],"performs":[106],"standard":[107],"decoding.":[108],"This":[109],"decouples":[110],"privacy":[112,123],"budget":[113],"from":[114],"generation,":[115],"enabling":[116],"arbitrarily":[117],"synthetic":[119],"samples":[120],"without":[121],"additional":[122],"cost":[124],"yielding":[126],"strong":[127],"fidelity":[128],"even":[129],"low-data":[131,169],"regimes.":[132],"Furthermore,":[133],"we":[134],"enhance":[135],"our":[136],"method":[137],"by":[138],"utilizing":[139],"pretrained":[140],"(base)":[141],"introducing":[144],"fixed-shot":[145],"prompting":[146],"boost":[148],"diversity":[150],"fidelity.":[152],"Our":[153],"experiments":[154],"demonstrate":[155],"outperforms":[158],"baselines":[160],"alignment":[163],"utility,":[166],"particularly":[167],"regimes,":[170],"while":[171],"significantly":[172],"reducing":[173],"computational":[174],"overhead.":[175]},"counts_by_year":[],"updated_date":"2026-07-15T18:14:33.161393","created_date":"2026-02-27T00:00:00"}