{"id":"https://openalex.org/W7154509587","doi":"https://doi.org/10.48550/arxiv.2604.12780","title":"Efficient Adversarial Training via Criticality-Aware Fine-Tuning","display_name":"Efficient Adversarial Training via Criticality-Aware Fine-Tuning","publication_year":2026,"publication_date":"2026-04-14","ids":{"openalex":"https://openalex.org/W7154509587","doi":"https://doi.org/10.48550/arxiv.2604.12780"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.12780","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.12780","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2604.12780","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5133632974","display_name":"Wenyun Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Wenyun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100459087","display_name":"Zheng Zhang","orcid":"https://orcid.org/0000-0002-0062-517X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Zheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133704134","display_name":"Dongmei Jiang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiang, Dongmei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133662809","display_name":"Yaowei Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Yaowei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5133643260","display_name":"Xiangyuan Lan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lan, Xiangyuan","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":null,"has_fulltext":false,"cited_by_count":0,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.9182000160217285,"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":0.9182000160217285,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.019700000062584877,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T10036","display_name":"Advanced Neural Network Applications","score":0.018699999898672104,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/adversarial-system","display_name":"Adversarial system","score":0.9093000292778015},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.7505000233650208},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.7498000264167786},{"id":"https://openalex.org/keywords/generalization","display_name":"Generalization","score":0.48739999532699585},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.4742000102996826},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.4293000102043152}],"concepts":[{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.9093000292778015},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.7505000233650208},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.7498000264167786},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6621999740600586},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.589900016784668},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5121999979019165},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.48739999532699585},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.4742000102996826},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.4293000102043152},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.34389999508857727},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.28459998965263367},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.27720001339912415},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.2680000066757202},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2531999945640564}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.12780","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.12780","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.12780","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.12780","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Vision":[0],"Transformer":[1],"(ViT)":[2],"models":[3,27,44],"have":[4],"achieved":[5],"remarkable":[6],"performance":[7],"across":[8],"various":[9],"vision":[10,167],"tasks,":[11],"with":[12,181,218],"scalability":[13,24],"being":[14],"a":[15,90,111,156,188],"key":[16],"advantage":[17],"when":[18,163],"applied":[19],"to":[20,28,45,71,86,95,99,118,135,144,165],"large":[21,78],"datasets.":[22],"This":[23],"enables":[25],"ViT":[26,43,79],"exhibit":[29],"strong":[30],"generalization":[31,162],"capabilities.":[32],"However,":[33],"as":[34],"the":[35,40,57,67,119,150,172],"number":[36,151],"of":[37,42,56,93,152,198],"parameters":[38,94,131,154],"increases,":[39],"robustness":[41,97,193],"adversarial":[46,136,175,183,192,207],"examples":[47],"does":[48],"not":[49],"scale":[50],"proportionally.":[51],"Adversarial":[52,108],"training":[53,176],"(AT),":[54],"one":[55],"most":[58,120,134],"effective":[59],"methods":[60,217],"for":[61,77,174],"enhancing":[62],"robustness,":[63],"typically":[64],"requires":[65],"fine-tuning":[66,123,142],"entire":[68],"model,":[69],"leading":[70],"prohibitively":[72],"high":[73],"computational":[74],"costs,":[75],"especially":[76],"architectures.":[80],"In":[81],"this":[82],"paper,":[83],"we":[84,105],"aim":[85],"robustly":[87,145],"fine-tune":[88],"only":[89,124,187],"small":[91],"subset":[92],"achieve":[96],"comparable":[98],"standard":[100],"AT.":[101],"To":[102],"accomplish":[103],"this,":[104],"introduce":[106],"Criticality-Aware":[107],"Training":[109],"(CAAT),":[110],"novel":[112],"method":[113],"that":[114,132,211],"adaptively":[115],"allocates":[116],"resources":[117],"robustness-critical":[121],"parameters,":[122],"selected":[125],"modules.":[126],"Specifically,":[127],"CAAT":[128,159,185,212],"efficiently":[129],"identifies":[130],"contribute":[133],"robustness.":[137],"It":[138],"then":[139],"leverages":[140],"parameter-efficient":[141],"(PEFT)":[143],"adjust":[146],"weight":[147],"matrices":[148],"where":[149],"critical":[153],"exceeds":[155],"predefined":[157],"threshold.":[158],"exhibits":[160],"favorable":[161],"scaled":[164],"larger":[166],"transformer":[168],"architectures,":[169],"potentially":[170],"paving":[171],"way":[173],"at":[177],"scale,":[178],"e.g,":[179],"compared":[180],"plain":[182],"training,":[184],"incurs":[186],"4.3%":[189],"decrease":[190],"in":[191],"while":[194],"tuning":[195],"approximately":[196],"6%":[197],"its":[199],"parameters.":[200,221],"Extensive":[201],"experiments":[202],"on":[203],"three":[204],"widely":[205],"used":[206],"learning":[208],"datasets":[209],"demonstrate":[210],"outperforms":[213],"state-of-the-art":[214],"lightweight":[215],"AT":[216],"fewer":[219],"trainable":[220]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-04-16T00:00:00"}