{"id":"https://openalex.org/W4402670002","doi":"https://doi.org/10.18653/v1/2024.findings-acl.468","title":"UOR: Universal Backdoor Attacks on Pre-trained Language Models","display_name":"UOR: Universal Backdoor Attacks on Pre-trained Language Models","publication_year":2024,"publication_date":"2024-01-01","ids":{"openalex":"https://openalex.org/W4402670002","doi":"https://doi.org/10.18653/v1/2024.findings-acl.468"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2024.findings-acl.468","doi":"https://doi.org/10.18653/v1/2024.findings-acl.468","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2024.findings-acl.468","pdf_url":"https://aclanthology.org/2024.findings-acl.468.pdf","source":{"id":"https://openalex.org/S4363605144","display_name":"Findings of the Association for Computational Linguistics: ACL (Findings of ACL)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics ACL 2024","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","is_traditional":true,"indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://aclanthology.org/2024.findings-acl.468.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Wei Du","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wei Du","raw_affiliation_strings":["University , School of Cyber Science and Engineering"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University , School of Cyber Science and Engineering","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103214488","display_name":"Peixuan Li","orcid":"https://orcid.org/0009-0005-9668-6282","observed_orcids":["https://orcid.org/0009-0005-9668-6282"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Peixuan Li","raw_affiliation_strings":["University , School of Cyber Science and Engineering"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University , School of Cyber Science and Engineering","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5025780527","display_name":"Haodong Zhao","orcid":"https://orcid.org/0000-0002-4405-1649","observed_orcids":["https://orcid.org/0000-0002-4405-1649"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Haodong Zhao","raw_affiliation_strings":["University , School of Cyber Science and Engineering"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University , School of Cyber Science and Engineering","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5035829367","display_name":"Tianjie Ju","orcid":"https://orcid.org/0009-0006-6978-1935","observed_orcids":["https://orcid.org/0009-0006-6978-1935"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tianjie Ju","raw_affiliation_strings":["University , School of Cyber Science and Engineering"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University , School of Cyber Science and Engineering","institution_ids":[]}]},{"author_position":"middle","author":{"id":null,"display_name":"Ge Ren","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ge Ren","raw_affiliation_strings":["University , School of Cyber Science and Engineering"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University , School of Cyber Science and Engineering","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5085695760","display_name":"Gongshen Liu","orcid":"https://orcid.org/0000-0001-5194-1570","observed_orcids":["https://orcid.org/0000-0001-5194-1570"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gongshen Liu","raw_affiliation_strings":["University , School of Cyber Science and Engineering"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University , School of Cyber Science and Engineering","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":4.4201,"has_fulltext":true,"cited_by_count":31,"citation_normalized_percentile":{"value":0.95943622,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":95,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"7865","last_page":"7877"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":1.0,"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":1.0,"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":0.0,"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/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.0,"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/backdoor-attacks","display_name":"backdoor attacks","score":0.9710000157356262},{"id":"https://openalex.org/keywords/pre-trained-language-models","display_name":"pre-trained language models","score":0.9679999947547913},{"id":"https://openalex.org/keywords/trigger-selection","display_name":"trigger selection","score":0.5830000042915344},{"id":"https://openalex.org/keywords/gradient-based-search","display_name":"gradient search","score":0.8999999761581421},{"id":"https://openalex.org/keywords/text-classification","display_name":"text classification","score":0.7049999833106995},{"id":"https://openalex.org/keywords/model-security","display_name":"model security","score":0.5260000228881836},{"id":"https://openalex.org/keywords/adversarial-attacks","display_name":"adversarial attacks","score":0.39500001072883606}],"concepts":[{"id":"https://openalex.org/C2781045450","wikidata":"https://www.wikidata.org/wiki/Q254569","display_name":"Backdoor","level":2,"score":0.9936079978942871},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7180193662643433},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3762145936489105},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.30863672494888306}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2024.findings-acl.468","doi":"https://doi.org/10.18653/v1/2024.findings-acl.468","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2024.findings-acl.468","pdf_url":"https://aclanthology.org/2024.findings-acl.468.pdf","source":{"id":"https://openalex.org/S4363605144","display_name":"Findings of the Association for Computational Linguistics: ACL (Findings of ACL)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics ACL 2024","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2305.09574","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2305.09574","pdf_url":"https://arxiv.org/pdf/2305.09574","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2305.09574","doi":"https://doi.org/10.48550/arxiv.2305.09574","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2305.09574","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":"doi:10.18653/v1/2024.findings-acl.468","doi":"https://doi.org/10.18653/v1/2024.findings-acl.468","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2024.findings-acl.468","pdf_url":"https://aclanthology.org/2024.findings-acl.468.pdf","source":{"id":"https://openalex.org/S4363605144","display_name":"Findings of the Association for Computational Linguistics: ACL (Findings of ACL)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics ACL 2024","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G7272630985","display_name":null,"funder_award_id":"U21B2020","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4402670002.pdf","grobid_xml":"https://content.openalex.org/works/W4402670002.grobid-xml"},"referenced_works_count":15,"referenced_works":["https://openalex.org/W2525332836","https://openalex.org/W2890727387","https://openalex.org/W2896457183","https://openalex.org/W2970290563","https://openalex.org/W2982399380","https://openalex.org/W3035367371","https://openalex.org/W3109409894","https://openalex.org/W3139080614","https://openalex.org/W3152956381","https://openalex.org/W3167002899","https://openalex.org/W3175052694","https://openalex.org/W3196832521","https://openalex.org/W3202492633","https://openalex.org/W3210951978","https://openalex.org/W4307076968"],"related_works":["https://openalex.org/W7111128812","https://openalex.org/W3210951978","https://openalex.org/W4411113104","https://openalex.org/W3202492633","https://openalex.org/W7170674412","https://openalex.org/W7133124865","https://openalex.org/W7166869350","https://openalex.org/W4385571902","https://openalex.org/W4401213182","https://openalex.org/W4416042731"],"abstract_inverted_index":{"Task-agnostic":[0],"and":[1,32,38,80,118,145],"transferable":[2],"backdoors":[3],"implanted":[4],"in":[5,97],"pre-trained":[6],"language":[7],"models":[8],"(PLMs)":[9],"pose":[10],"a":[11,50],"severe":[12],"security":[13],"threat":[14],"as":[15],"they":[16],"can":[17,75,112],"be":[18,113],"inherited":[19],"to":[20,106,115,133],"any":[21],"downstream":[22,98],"task.However,":[23],"existing":[24],"methods":[25],"rely":[26],"on":[27,127,138],"manual":[28,63,134],"selection":[29,64],"of":[30,89],"triggers":[31],"backdoor":[33,52,82],"representations,":[34],"hindering":[35],"their":[36],"effectiveness":[37],"universality":[39],"across":[40],"different":[41,116,141],"PLMs":[42,117,139],"or":[43],"usage":[44,143],"paradigms.In":[45],"this":[46],"paper,":[47],"we":[48,68,102,136],"propose":[49],"new":[51],"attack":[53,125],"method":[54],"called":[55],"UOR,":[56],"which":[57,74],"overcomes":[58],"these":[59],"limitations":[60],"by":[61],"turning":[62],"into":[65],"automatic":[66],"optimization.Specifically,":[67],"design":[69],"poisoned":[70],"supervised":[71],"contrastive":[72],"learning,":[73],"automatically":[76],"learn":[77],"more":[78,86,95,146],"uniform":[79],"universal":[81],"representations.This":[83],"allows":[84],"for":[85,152],"even":[87],"coverage":[88],"the":[90],"output":[91],"space,":[92],"thus":[93],"hitting":[94],"labels":[96],"tasks":[99,131],"after":[100],"finetuning.Furthermore,":[101],"utilize":[103],"gradient":[104],"search":[105],"select":[107],"appropriate":[108],"trigger":[109],"words":[110],"that":[111,121],"adapted":[114],"vocabularies.Experiments":[119],"show":[120],"UOR":[122],"achieves":[123],"better":[124],"performance":[126],"various":[128],"text":[129],"classification":[130],"compared":[132],"methods.Moreover,":[135],"test":[137],"with":[140],"architectures,":[142],"paradigms,":[144],"challenging":[147],"tasks,":[148],"achieving":[149],"higher":[150],"scores":[151],"universality.":[153]},"counts_by_year":[{"year":2026,"cited_by_count":6},{"year":2025,"cited_by_count":13},{"year":2024,"cited_by_count":9},{"year":2023,"cited_by_count":3}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2024-09-21T00:00:00"}