{"id":"https://openalex.org/W4401751543","doi":"https://doi.org/10.1109/icccn61486.2024.10637573","title":"Agent Deprogramming Finetuning Away Backdoor Triggers for Secure Machine Learning Models/LLMs","display_name":"Agent Deprogramming Finetuning Away Backdoor Triggers for Secure Machine Learning Models/LLMs","publication_year":2024,"publication_date":"2024-07-29","ids":{"openalex":"https://openalex.org/W4401751543","doi":"https://doi.org/10.1109/icccn61486.2024.10637573"},"language":"en","primary_location":{"id":"doi:10.1109/icccn61486.2024.10637573","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1109/icccn61486.2024.10637573","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2024 33rd International Conference on Computer Communications and Networks (ICCCN)","raw_type":"proceedings-article"},"type":"conference-paper","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":"https://openalex.org/A5057931433","display_name":"Demarcus Edwards","orcid":null},"institutions":[{"id":"https://openalex.org/I137853757","display_name":"Howard University","ror":"https://ror.org/05gt1vc06","country_code":"US","type":"education","lineage":["https://openalex.org/I137853757"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Demarcus Edwards","raw_affiliation_strings":["Howard University,Department of Electrical Engineering and Computer Science,Washington,DC,USA,20059"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Howard University,Department of Electrical Engineering and Computer Science,Washington,DC,USA,20059","institution_ids":["https://openalex.org/I137853757"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5046343080","display_name":"Danda B. Rawat","orcid":"https://orcid.org/0000-0003-3638-3464"},"institutions":[{"id":"https://openalex.org/I137853757","display_name":"Howard University","ror":"https://ror.org/05gt1vc06","country_code":"US","type":"education","lineage":["https://openalex.org/I137853757"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Danda B. Rawat","raw_affiliation_strings":["Howard University,Department of Electrical Engineering and Computer Science,Washington,DC,USA,20059"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Howard University,Department of Electrical Engineering and Computer Science,Washington,DC,USA,20059","institution_ids":["https://openalex.org/I137853757"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I137853757"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"6"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10400","display_name":"Network Security and Intrusion Detection","score":0.7908999919891357,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T10400","display_name":"Network Security and Intrusion Detection","score":0.7908999919891357,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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","display_name":"Backdoor","score":0.9540125131607056},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6320302486419678},{"id":"https://openalex.org/keywords/computer-security","display_name":"Computer security","score":0.49106407165527344},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3809923827648163}],"concepts":[{"id":"https://openalex.org/C2781045450","wikidata":"https://www.wikidata.org/wiki/Q254569","display_name":"Backdoor","level":2,"score":0.9540125131607056},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6320302486419678},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.49106407165527344},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3809923827648163}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icccn61486.2024.10637573","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1109/icccn61486.2024.10637573","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2024 33rd International Conference on Computer Communications and Networks (ICCCN)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320313488","display_name":"Howard University","ror":"https://ror.org/05gt1vc06"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":14,"referenced_works":["https://openalex.org/W2774423163","https://openalex.org/W3034258347","https://openalex.org/W3035367371","https://openalex.org/W3121478722","https://openalex.org/W3168867926","https://openalex.org/W4226278401","https://openalex.org/W4296413526","https://openalex.org/W4385894687","https://openalex.org/W4387635776","https://openalex.org/W4390833079","https://openalex.org/W4402264526","https://openalex.org/W6782879696","https://openalex.org/W6796581206","https://openalex.org/W6810738896"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2748952813","https://openalex.org/W4320031223","https://openalex.org/W4200629851","https://openalex.org/W4281902577","https://openalex.org/W4309417370","https://openalex.org/W4292107232","https://openalex.org/W3009072493","https://openalex.org/W4386080799","https://openalex.org/W3140988292"],"abstract_inverted_index":{"Adversarial":[0],"fine-tuning":[1,31,40,44,84,92,106,135],"has":[2],"shown":[3],"promise":[4],"in":[5,15,19,51,146],"improving":[6],"model":[7],"robustness":[8],"against":[9],"test-time":[10],"perturbations,":[11],"but":[12],"its":[13],"effectiveness":[14,57],"mitigating":[16],"backdoor":[17,49,110,144],"attacks":[18,50,145],"language":[20,52,147],"models":[21,67,99],"remains":[22],"understudied.":[23],"This":[24],"work":[25],"investigates":[26],"the":[27,56,63,71,74,80,90,140],"use":[28],"of":[29,58,65,73,82,143],"adversarial":[30,134],"techniques":[32,60,125],"and":[33,43,69,86,123],"deprogramming,":[34],"specifically":[35],"focusing":[36],"on":[37],"basic":[38],"fine-tuning,":[39],"with":[41,45,68,121,136],"paraphrasing,":[42],"retokenization,":[46],"to":[47,95,112],"mitigate":[48],"models.":[53,148],"We":[54],"evaluate":[55],"these":[59],"by":[61],"measuring":[62],"accuracy":[64],"our":[66,98,115],"without":[70],"context":[72],"original":[75],"query.":[76],"Our":[77],"approach":[78],"extends":[79],"idea":[81],"supervised":[83,105],"(SFT)":[85],"incorporates":[87],"adaptors":[88],"from":[89],"parameter-efficient":[91],"(PEFT)":[93],"framework":[94],"efficiently":[96],"fine-tune":[97],"while":[100],"minimizing":[101],"computational":[102],"resources.":[103],"While":[104],"alone":[107],"can":[108,126],"alleviate":[109],"triggers":[111],"some":[113],"extent,":[114],"results":[116],"demonstrate":[117],"that":[118],"combining":[119],"SFT":[120],"paraphrasing":[122,137],"retokenization":[124],"achieve":[127],"even":[128],"more":[129],"effective":[130],"mitigation.":[131],"In":[132],"particular,":[133],"significantly":[138],"reduces":[139],"success":[141],"rate":[142]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}