{"id":"https://openalex.org/W4415593644","doi":"https://doi.org/10.1109/tcsvt.2025.3625396","title":"Evaluating the Adversarial Robustness of Vision-Language Models via Internal Feature Perturbations","display_name":"Evaluating the Adversarial Robustness of Vision-Language Models via Internal Feature Perturbations","publication_year":2025,"publication_date":"2025-10-27","ids":{"openalex":"https://openalex.org/W4415593644","doi":"https://doi.org/10.1109/tcsvt.2025.3625396"},"language":null,"primary_location":{"id":"doi:10.1109/tcsvt.2025.3625396","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcsvt.2025.3625396","pdf_url":null,"source":{"id":"https://openalex.org/S115173108","display_name":"IEEE Transactions on Circuits and Systems for Video Technology","issn_l":"1051-8215","issn":["1051-8215","1558-2205"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Circuits and Systems for Video Technology","raw_type":"journal-article"},"type":"article","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/A5061846588","display_name":"Chaohu Liu","orcid":"https://orcid.org/0009-0001-7588-4264"},"institutions":[{"id":"https://openalex.org/I126520041","display_name":"University of Science and Technology of China","ror":"https://ror.org/04c4dkn09","country_code":"CN","type":"education","lineage":["https://openalex.org/I126520041","https://openalex.org/I19820366"]},{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chaohu Liu","raw_affiliation_strings":["Tencent YouTu Lab, Shenzhen, China","School of Computer Science and Technology, University of Science and Technology of China, Hefei, China"],"raw_orcid":"https://orcid.org/0009-0001-7588-4264","affiliations":[{"raw_affiliation_string":"Tencent YouTu Lab, Shenzhen, China","institution_ids":["https://openalex.org/I2250653659"]},{"raw_affiliation_string":"School of Computer Science and Technology, University of Science and Technology of China, Hefei, China","institution_ids":["https://openalex.org/I126520041"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5107959424","display_name":"Yubo Wang","orcid":"https://orcid.org/0009-0000-7454-1019"},"institutions":[{"id":"https://openalex.org/I126520041","display_name":"University of Science and Technology of China","ror":"https://ror.org/04c4dkn09","country_code":"CN","type":"education","lineage":["https://openalex.org/I126520041","https://openalex.org/I19820366"]},{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yubo Wang","raw_affiliation_strings":["Tencent YouTu Lab, Shenzhen, China","School of Artificial Intelligence and Data Science, University of Science and Technology of China, Hefei, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tencent YouTu Lab, Shenzhen, China","institution_ids":["https://openalex.org/I2250653659"]},{"raw_affiliation_string":"School of Artificial Intelligence and Data Science, University of Science and Technology of China, Hefei, China","institution_ids":["https://openalex.org/I126520041"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5068313625","display_name":"Haoyu Cao","orcid":"https://orcid.org/0000-0002-3789-9705"},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Haoyu Cao","raw_affiliation_strings":["Tencent YouTu Lab, Shenzhen, China","Tencent YouTu Lab, China"],"raw_orcid":"https://orcid.org/0000-0002-3789-9705","affiliations":[{"raw_affiliation_string":"Tencent YouTu Lab, Shenzhen, China","institution_ids":["https://openalex.org/I2250653659"]},{"raw_affiliation_string":"Tencent YouTu Lab, China","institution_ids":["https://openalex.org/I2250653659"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100339921","display_name":"Bing Liu","orcid":"https://orcid.org/0000-0002-2365-6606"},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Bing Liu","raw_affiliation_strings":["Tencent YouTu Lab, Shenzhen, China","Tencent YouTu Lab, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tencent YouTu Lab, Shenzhen, China","institution_ids":["https://openalex.org/I2250653659"]},{"raw_affiliation_string":"Tencent YouTu Lab, China","institution_ids":["https://openalex.org/I2250653659"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5016912950","display_name":"Deqiang Jiang","orcid":null},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Deqiang Jiang","raw_affiliation_strings":["Tencent YouTu Lab, Shenzhen, China","Tencent YouTu Lab, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Tencent YouTu Lab, Shenzhen, China","institution_ids":["https://openalex.org/I2250653659"]},{"raw_affiliation_string":"Tencent YouTu Lab, China","institution_ids":["https://openalex.org/I2250653659"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":1.2699,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":{"value":0.86214083,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":94,"max":97},"biblio":{"volume":"36","issue":"3","first_page":"3938","last_page":"3950"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.9991000294685364,"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":0.9991000294685364,"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/T11512","display_name":"Anomaly Detection Techniques and Applications","score":0.9172000288963318,"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":0.7414000034332275},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.7239000201225281},{"id":"https://openalex.org/keywords/entropy","display_name":"Entropy (arrow of time)","score":0.6186000108718872},{"id":"https://openalex.org/keywords/limiting","display_name":"Limiting","score":0.4871000051498413},{"id":"https://openalex.org/keywords/exploit","display_name":"Exploit","score":0.40709999203681946},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.3580999970436096},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.3490999937057495},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.33239999413490295}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7760999798774719},{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.7414000034332275},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.7239000201225281},{"id":"https://openalex.org/C106301342","wikidata":"https://www.wikidata.org/wiki/Q4117933","display_name":"Entropy (arrow of time)","level":2,"score":0.6186000108718872},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5019999742507935},{"id":"https://openalex.org/C188198153","wikidata":"https://www.wikidata.org/wiki/Q1613840","display_name":"Limiting","level":2,"score":0.4871000051498413},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4366999864578247},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.40709999203681946},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.3580999970436096},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.3490999937057495},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.34040001034736633},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.33239999413490295},{"id":"https://openalex.org/C171752962","wikidata":"https://www.wikidata.org/wiki/Q255166","display_name":"Kullback\u2013Leibler divergence","level":2,"score":0.32409998774528503},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.32179999351501465},{"id":"https://openalex.org/C177918212","wikidata":"https://www.wikidata.org/wiki/Q803623","display_name":"Perturbation (astronomy)","level":2,"score":0.3197999894618988},{"id":"https://openalex.org/C2984842247","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep neural networks","level":3,"score":0.31209999322891235},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.2935999929904938},{"id":"https://openalex.org/C167981619","wikidata":"https://www.wikidata.org/wiki/Q1685498","display_name":"Cross entropy","level":3,"score":0.29010000824928284},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.2897999882698059},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2854999899864197},{"id":"https://openalex.org/C52622490","wikidata":"https://www.wikidata.org/wiki/Q1026626","display_name":"Feature extraction","level":2,"score":0.2721000015735626},{"id":"https://openalex.org/C182049051","wikidata":"https://www.wikidata.org/wiki/Q17147155","display_name":"Transfer entropy","level":3,"score":0.26649999618530273},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.26350000500679016},{"id":"https://openalex.org/C52622258","wikidata":"https://www.wikidata.org/wiki/Q131222","display_name":"Information theory","level":2,"score":0.2597000002861023},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2565000057220459},{"id":"https://openalex.org/C9679016","wikidata":"https://www.wikidata.org/wiki/Q1417473","display_name":"Principle of maximum entropy","level":2,"score":0.25540000200271606}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tcsvt.2025.3625396","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tcsvt.2025.3625396","pdf_url":null,"source":{"id":"https://openalex.org/S115173108","display_name":"IEEE Transactions on Circuits and Systems for Video Technology","issn_l":"1051-8215","issn":["1051-8215","1558-2205"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Circuits and Systems for Video Technology","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":42,"referenced_works":["https://openalex.org/W1861492603","https://openalex.org/W2108598243","https://openalex.org/W2774644650","https://openalex.org/W2963542245","https://openalex.org/W3011813266","https://openalex.org/W3035031253","https://openalex.org/W3061797349","https://openalex.org/W3135970545","https://openalex.org/W3143373604","https://openalex.org/W3207117534","https://openalex.org/W4312233877","https://openalex.org/W4312790346","https://openalex.org/W4312820932","https://openalex.org/W4376607895","https://openalex.org/W4385245566","https://openalex.org/W4386038437","https://openalex.org/W4386066003","https://openalex.org/W4386071707","https://openalex.org/W4386072305","https://openalex.org/W4390190425","https://openalex.org/W4390872436","https://openalex.org/W4390874708","https://openalex.org/W4393157467","https://openalex.org/W4396982312","https://openalex.org/W4402704527","https://openalex.org/W4402716330","https://openalex.org/W4402727038","https://openalex.org/W4402727764","https://openalex.org/W4404035240","https://openalex.org/W4404356490","https://openalex.org/W4404725774","https://openalex.org/W4406110392","https://openalex.org/W4407097622","https://openalex.org/W4409129027","https://openalex.org/W4409724265","https://openalex.org/W4409869032","https://openalex.org/W4411948665","https://openalex.org/W4413639576","https://openalex.org/W4415795395","https://openalex.org/W4415796110","https://openalex.org/W7133193597","https://openalex.org/W7133254726"],"related_works":[],"abstract_inverted_index":{"Vision-language":[0],"models":[1,65],"(VLMs),":[2],"such":[3],"as":[4],"BLIP-2":[5],"and":[6,175],"LLaVA,":[7],"have":[8],"significantly":[9],"advanced":[10],"multimodal":[11],"understanding":[12],"but":[13,115],"exhibit":[14],"critical":[15],"vulnerabilities":[16],"to":[17,52,162],"visual":[18,79],"adversarial":[19,40,194],"perturbations.":[20],"The":[21],"high":[22],"efficacy":[23,102],"of":[24,60,110,164],"untargeted":[25,89],"attacks,":[26],"in":[27,76,196],"particular,":[28],"poses":[29],"significant":[30],"concerns":[31],"for":[32,192],"their":[33],"operational":[34],"robustness.":[35],"Conventional":[36],"attack":[37,90,101],"methods":[38],"generate":[39],"examples":[41],"by":[42,105,117,155],"backpropagating":[43],"the":[44,49,53,57,61,73,78,107,111,131,140,152],"language":[45,64],"modeling":[46],"loss":[47],"from":[48],"final":[50,113,141],"output":[51,114,142],"input":[54],"image.":[55],"However,":[56],"deep":[58],"architecture":[59],"integrated":[62],"large":[63],"(LLMs)":[66],"often":[67],"diminishes":[68],"this":[69,83],"gradient":[70],"flow,":[71],"limiting":[72],"attack\u2019s":[74],"effectiveness":[75],"perturbing":[77],"domain.":[80],"To":[81],"address":[82],"limitation,":[84],"we":[85],"introduce":[86],"a":[87,147,187],"novel":[88],"method":[91,180],"based":[92],"on":[93,139],"Maximizing":[94],"Information":[95],"Entropy":[96],"(MIE).":[97],"Our":[98],"approach":[99],"enhances":[100],"not":[103],"only":[104],"maximizing":[106],"information":[108],"entropy":[109],"model\u2019s":[112,132],"also":[116],"directly":[118],"inducing":[119],"uncertainty":[120,153],"within":[121],"its":[122],"internal":[123],"feature":[124],"representations.":[125],"This":[126],"layer-wise":[127],"perturbation":[128],"strategy":[129],"disrupts":[130],"cognitive":[133],"process":[134],"more":[135,189],"comprehensively":[136],"than":[137,159],"relying":[138],"layer":[143],"alone.":[144],"We":[145],"provide":[146],"theoretical":[148],"analysis":[149],"demonstrating":[150],"that":[151,163,178],"induced":[154],"MIE":[156],"is":[157],"greater":[158],"or":[160],"equal":[161],"conventional":[165],"output-only":[166],"attacks.":[167],"Comprehensive":[168],"quantitative":[169],"evaluations":[170],"across":[171],"multiple":[172],"VLM":[173],"architectures":[174],"datasets":[176],"confirm":[177],"our":[179],"consistently":[181],"outperforms":[182],"existing":[183],"techniques,":[184],"thereby":[185],"establishing":[186],"new,":[188],"rigorous":[190],"benchmark":[191],"assessing":[193],"robustness":[195],"vision-language":[197],"models.":[198]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-07-22T07:51:19.307946","created_date":"2025-10-28T00:00:00"}