{"id":"https://openalex.org/W7161748427","doi":"https://doi.org/10.48550/arxiv.2605.17577","title":"TAME: Test-Time Adversarial Prompt Tuning via Mixture-of-Experts for Vision-Language Models","display_name":"TAME: Test-Time Adversarial Prompt Tuning via Mixture-of-Experts for Vision-Language Models","publication_year":2026,"publication_date":"2026-05-17","ids":{"openalex":"https://openalex.org/W7161748427","doi":"https://doi.org/10.48550/arxiv.2605.17577"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.17577","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.17577","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.2605.17577","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5136474653","display_name":"Xin Wang","orcid":"https://orcid.org/0009-0004-3223-3228"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Xin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136472232","display_name":"Yixu Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Yixu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136502146","display_name":"Jiaming Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Jiaming","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136502330","display_name":"Ruofan Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Ruofan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136503459","display_name":"Jiaqi Yu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yu, Jiaqi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136499178","display_name":"Kai Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Kai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136499690","display_name":"Jingjing Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Jingjing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136475490","display_name":"Xingjun Ma","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ma, Xingjun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5136472197","display_name":"Yu-Gang Jiang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiang, Yu-Gang","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":0.9157000184059143,"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":0.9157000184059143,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.023000000044703484,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.012500000186264515,"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/adversarial-system","display_name":"Adversarial system","score":0.8669000267982483},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.6523000001907349},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6096000075340271},{"id":"https://openalex.org/keywords/entropy","display_name":"Entropy (arrow of time)","score":0.4731000065803528},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.385699987411499},{"id":"https://openalex.org/keywords/regularization","display_name":"Regularization (linguistics)","score":0.36629998683929443}],"concepts":[{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.8669000267982483},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7182999849319458},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.6523000001907349},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6096000075340271},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5598000288009644},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.552299976348877},{"id":"https://openalex.org/C106301342","wikidata":"https://www.wikidata.org/wiki/Q4117933","display_name":"Entropy (arrow of time)","level":2,"score":0.4731000065803528},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.385699987411499},{"id":"https://openalex.org/C2776135515","wikidata":"https://www.wikidata.org/wiki/Q17143721","display_name":"Regularization (linguistics)","level":2,"score":0.36629998683929443},{"id":"https://openalex.org/C2984842247","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep neural networks","level":3,"score":0.3476000130176544},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.2937999963760376},{"id":"https://openalex.org/C167981619","wikidata":"https://www.wikidata.org/wiki/Q1685498","display_name":"Cross entropy","level":3,"score":0.26759999990463257},{"id":"https://openalex.org/C2778403875","wikidata":"https://www.wikidata.org/wiki/Q20312394","display_name":"Adversarial machine learning","level":3,"score":0.2639999985694885}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.17577","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.17577","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.2605.17577","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.17577","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":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large-scale":[0],"pre-trained":[1],"Vision-Language":[2],"models":[3],"(VLMs),":[4],"such":[5],"as":[6],"CLIP,":[7],"exhibit":[8],"strong":[9],"zero-shot":[10,155,164],"generalization,":[11],"yet":[12],"remain":[13],"highly":[14],"vulnerable":[15],"to":[16,89,125],"imperceptible":[17],"adversarial":[18,129,165,189],"perturbations,":[19],"raising":[20],"serious":[21],"safety":[22],"concerns":[23],"for":[24,95,136],"open-world":[25],"deployment.":[26],"To":[27],"enhance":[28],"robustness":[29,166,200],"without":[30],"requiring":[31],"downstream":[32],"task-specific":[33],"retraining,":[34],"we":[35],"propose":[36],"TAME,":[37],"a":[38,77,91],"novel":[39],"test-time":[40,104],"defense.":[41,73],"Building":[42],"upon":[43],"our":[44],"prior":[45],"Test-Time":[46],"Adversarial":[47],"Prompt":[48],"Tuning":[49],"(TAPT),":[50],"TAME":[51,75,145,161,184],"introduces":[52],"an":[53,63,85,198],"architectural":[54],"reformulation":[55],"by":[56,109,171],"replacing":[57],"TAPT's":[58],"single":[59],"adaptive":[60,72],"prompt":[61,93,141,190,195],"with":[62],"input-conditioned":[64],"Mixture-of-Experts":[65],"(MoE)":[66],"framework,":[67],"enabling":[68],"more":[69],"expressive":[70],"and":[71,83,128,132,140,152],"Specifically,":[74],"maintains":[76],"bank":[78],"of":[79,121,167,202],"learnable":[80],"expert":[81,138],"prompts":[82],"employs":[84],"input-dependent":[86],"routing":[87],"mechanism":[88,106],"aggregate":[90],"customized":[92],"mixture":[94],"each":[96],"unlabeled":[97],"test":[98],"sample":[99],"at":[100,172,203],"inference":[101],"time.":[102],"This":[103],"defense":[105],"is":[107],"driven":[108],"three":[110],"unsupervised":[111],"objectives:":[112],"(1)":[113],"multi-view":[114],"prediction":[115],"entropy":[116],"minimization,":[117],"(2)":[118],"layer-wise":[119],"alignment":[120],"visual":[122],"token":[123],"statistics":[124],"precomputed":[126],"clean":[127,182],"reference":[130],"distributions,":[131],"(3)":[133],"MoE":[134],"regularization":[135],"balanced":[137],"utilization":[139],"diversity.":[142],"We":[143],"evaluated":[144],"on":[146,181],"11":[147],"benchmark":[148],"datasets,":[149],"including":[150],"ImageNet":[151],"10":[153],"additional":[154],"datasets.":[156],"The":[157],"results":[158],"show":[159],"that":[160],"improves":[162],"the":[163,168],"original":[169],"CLIP":[170],"least":[173,204],"49.1%":[174],"under":[175],"AutoAttack":[176],"while":[177],"largely":[178],"preserving":[179],"generalization":[180],"samples.":[183],"also":[185],"consistently":[186],"outperforms":[187],"existing":[188],"tuning":[191],"methods":[192],"across":[193],"multiple":[194],"designs,":[196],"yielding":[197],"average":[199],"gain":[201],"30.2%.":[205]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-20T00:00:00"}