{"id":"https://openalex.org/W7138133808","doi":"https://doi.org/10.1609/aaai.v40i40.40720","title":"PrivSV: Differentially Private Steering Vector for Large Language Models","display_name":"PrivSV: Differentially Private Steering Vector for Large Language Models","publication_year":2026,"publication_date":"2026-03-14","ids":{"openalex":"https://openalex.org/W7138133808","doi":"https://doi.org/10.1609/aaai.v40i40.40720"},"language":null,"primary_location":{"id":"doi:10.1609/aaai.v40i40.40720","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i40.40720","pdf_url":null,"source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"diamond","oa_url":"https://doi.org/10.1609/aaai.v40i40.40720","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5129685466","display_name":"Haocheng Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Haocheng Yang","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129673391","display_name":"Xiang Cheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiang Cheng","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129719872","display_name":"Chenhao Sun","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chenhao Sun","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129652182","display_name":"Pengfei Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pengfei Zhang","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5129645671","display_name":"Sen Su","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sen Su","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5129685466"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.35313002,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"40","issue":"40","first_page":"34241","last_page":"34249"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.8202000260353088,"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":0.8202000260353088,"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":0.014999999664723873,"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/T10028","display_name":"Topic Modeling","score":0.014100000262260437,"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/differential-privacy","display_name":"Differential privacy","score":0.6765999794006348},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.5824999809265137},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.5684000253677368},{"id":"https://openalex.org/keywords/compensation","display_name":"Compensation (psychology)","score":0.47870001196861267},{"id":"https://openalex.org/keywords/reduction","display_name":"Reduction (mathematics)","score":0.46230000257492065},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.44429999589920044},{"id":"https://openalex.org/keywords/measure","display_name":"Measure (data warehouse)","score":0.43529999256134033},{"id":"https://openalex.org/keywords/noise-reduction","display_name":"Noise reduction","score":0.4226999878883362}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7021999955177307},{"id":"https://openalex.org/C23130292","wikidata":"https://www.wikidata.org/wiki/Q5275358","display_name":"Differential privacy","level":2,"score":0.6765999794006348},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.5824999809265137},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.5684000253677368},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.49000000953674316},{"id":"https://openalex.org/C2780023022","wikidata":"https://www.wikidata.org/wiki/Q1338171","display_name":"Compensation (psychology)","level":2,"score":0.47870001196861267},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.46230000257492065},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.44429999589920044},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.43529999256134033},{"id":"https://openalex.org/C163294075","wikidata":"https://www.wikidata.org/wiki/Q581861","display_name":"Noise reduction","level":2,"score":0.4226999878883362},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3603000044822693},{"id":"https://openalex.org/C2780909371","wikidata":"https://www.wikidata.org/wiki/Q4801092","display_name":"Artificial noise","level":4,"score":0.3587000072002411},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.33959999680519104},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3359000086784363},{"id":"https://openalex.org/C93226319","wikidata":"https://www.wikidata.org/wiki/Q193137","display_name":"Differential (mechanical device)","level":2,"score":0.33250001072883606},{"id":"https://openalex.org/C12267149","wikidata":"https://www.wikidata.org/wiki/Q282453","display_name":"Support vector machine","level":2,"score":0.3301999866962433},{"id":"https://openalex.org/C123201435","wikidata":"https://www.wikidata.org/wiki/Q456632","display_name":"Information privacy","level":2,"score":0.3240000009536743},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3100999891757965},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2957000136375427},{"id":"https://openalex.org/C99221444","wikidata":"https://www.wikidata.org/wiki/Q1532069","display_name":"Private information retrieval","level":2,"score":0.28679999709129333},{"id":"https://openalex.org/C2778334786","wikidata":"https://www.wikidata.org/wiki/Q1586270","display_name":"Variation (astronomy)","level":2,"score":0.2741999924182892},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.263700008392334},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.251800000667572}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1609/aaai.v40i40.40720","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i40.40720","pdf_url":null,"source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1609/aaai.v40i40.40720","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i40.40720","pdf_url":null,"source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"sustainable_development_goals":[{"display_name":"Peace, Justice and strong institutions","score":0.46494153141975403,"id":"https://metadata.un.org/sdg/16"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Steering":[0],"Vector":[1],"(SV)":[2],"is":[3],"a":[4,66,93,107,121],"powerful":[5],"technique":[6],"for":[7,43,70],"controlling":[8],"Large":[9],"Language":[10],"Models":[11],"(LLMs)":[12],"by":[13,117],"manipulating":[14],"their":[15],"activations":[16],"without":[17],"altering":[18],"model":[19],"weights.":[20],"However,":[21],"when":[22],"constructed":[23],"from":[24],"sensitive":[25],"data,":[26],"SV":[27,45,52,72,79],"poses":[28],"significant":[29],"privacy":[30,40,161],"risks,":[31],"as":[32],"it":[33],"may":[34],"leak":[35],"private":[36],"information.":[37],"Existing":[38],"differential":[39],"(DP)":[41],"techniques":[42,158],"constructing":[44,71],"cannot":[46],"be":[47],"directly":[48],"applied":[49],"to":[50,99,113,128],"training-based":[51],"construction":[53,80],"paradigms,":[54],"which":[55],"offer":[56],"higher":[57],"task":[58],"performance.":[59],"In":[60,86],"this":[61],"work,":[62],"we":[63,88],"present":[64],"**PrivSV**,":[65],"general":[67],"privacy-preserving":[68],"approach":[69],"with":[73,77],"DP":[74],"guarantees,":[75],"compatible":[76],"arbitrary":[78],"paradigms":[81],"while":[82],"maintaining":[83],"high":[84],"utility.":[85],"PrivSV,":[87],"propose":[89],"three":[90],"novel":[91],"methods:":[92],"Layer-wise":[94],"Noise-Resilient":[95],"Reduction":[96],"(LNR\u00b2)":[97],"method":[98,112,127],"reduce":[100],"the":[101,131,134],"injected":[102],"noise":[103,118],"in":[104],"high-dimensional":[105],"SV;":[106],"Directional":[108],"Prior":[109],"Compensation":[110],"(DPC)":[111],"recover":[114],"utility":[115],"degraded":[116],"perturbation;":[119],"and":[120,150],"Privacy-Aware":[122],"Optimal":[123],"Parameter":[124],"Determination":[125],"(POPD)":[126],"adaptively":[129],"maximize":[130],"performance":[132],"of":[133,143],"final":[135],"compensated":[136],"SV.":[137],"Extensive":[138],"experiments":[139],"on":[140],"open-source":[141],"LLMs":[142],"different":[144],"families":[145],"(i.e.,":[146],"LlaMa,":[147],"Qwen,":[148],"Mistral":[149],"Gemma)":[151],"demonstrate":[152],"that":[153],"PrivSV":[154],"outperforms":[155],"several":[156],"existing":[157],"across":[159],"various":[160],"budgets.":[162]},"counts_by_year":[],"updated_date":"2026-03-18T06:31:55.123368","created_date":"2026-03-18T00:00:00"}