{"id":"https://openalex.org/W7225299095","doi":"https://doi.org/10.21203/rs.3.rs-11336583/v1","title":"FASDP: Frequency-Aware Scale-Diverse Purification against Pixel-Space Visual Jailbreaks","display_name":"FASDP: Frequency-Aware Scale-Diverse Purification against Pixel-Space Visual Jailbreaks","publication_year":2026,"publication_date":"2026-11-06","ids":{"openalex":"https://openalex.org/W7225299095","doi":"https://doi.org/10.21203/rs.3.rs-11336583/v1"},"language":"en","primary_location":{"id":"doi:10.21203/rs.3.rs-11336583/v1","doi":"https://doi.org/10.21203/rs.3.rs-11336583/v1","is_oa":true,"landing_page_url":"https://doi.org/10.21203/rs.3.rs-11336583/v1","pdf_url":"https://www.researchsquare.com/article/rs-11336583/latest.pdf","source":{"id":"https://openalex.org/S4306525896","display_name":"Research Square","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":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Research Square","raw_type":"posted-content"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://www.researchsquare.com/article/rs-11336583/latest.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5135360656","display_name":"Xiaolong Qin","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I24201400","display_name":"Chengdu University of Information Technology","ror":"https://ror.org/01yxwrh59","country_code":"CN","type":"education","lineage":["https://openalex.org/I24201400"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xiaolong Qin","raw_affiliation_strings":["Chengdu University of Information Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chengdu University of Information Technology","institution_ids":["https://openalex.org/I24201400"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5087468967","display_name":"Chunjiang Wu","orcid":"https://orcid.org/0009-0000-0576-9677","observed_orcids":["https://orcid.org/0009-0000-0576-9677"]},"institutions":[{"id":"https://openalex.org/I24201400","display_name":"Chengdu University of Information Technology","ror":"https://ror.org/01yxwrh59","country_code":"CN","type":"education","lineage":["https://openalex.org/I24201400"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chunjiang Wu","raw_affiliation_strings":["Chengdu University of Information Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chengdu University of Information Technology","institution_ids":["https://openalex.org/I24201400"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5160970044","display_name":"YiLe Li","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I24201400","display_name":"Chengdu University of Information Technology","ror":"https://ror.org/01yxwrh59","country_code":"CN","type":"education","lineage":["https://openalex.org/I24201400"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"YiLe Li","raw_affiliation_strings":["Chengdu University of Information Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chengdu University of Information Technology","institution_ids":["https://openalex.org/I24201400"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5160904034","display_name":"TingHui Chen","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I24201400","display_name":"Chengdu University of Information Technology","ror":"https://ror.org/01yxwrh59","country_code":"CN","type":"education","lineage":["https://openalex.org/I24201400"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"TingHui Chen","raw_affiliation_strings":["Chengdu University of Information Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chengdu University of Information Technology","institution_ids":["https://openalex.org/I24201400"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134697693","display_name":"Zhibin Li","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I24201400","display_name":"Chengdu University of Information Technology","ror":"https://ror.org/01yxwrh59","country_code":"CN","type":"education","lineage":["https://openalex.org/I24201400"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhibin Li","raw_affiliation_strings":["Chengdu University of Information Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chengdu University of Information Technology","institution_ids":["https://openalex.org/I24201400"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5123399124","display_name":"Shilin Qiu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I31595395","display_name":"Chengdu University of Technology","ror":"https://ror.org/05pejbw21","country_code":"CN","type":"education","lineage":["https://openalex.org/I31595395"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"ShiLin Qiu","raw_affiliation_strings":["Chengdu University of Technology"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chengdu University of Technology","institution_ids":["https://openalex.org/I31595395"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"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":0.9994999766349792,"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":0.9994999766349792,"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/T12357","display_name":"Digital Media Forensic Detection","score":0.00019999999494757503,"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/T11424","display_name":"Security and Verification in Computing","score":0.0,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/vision-language-models","display_name":"vision-language models","score":0.8420000076293945},{"id":"https://openalex.org/keywords/multi-scale-denoising","display_name":"multi-scale denoising","score":0.7110000252723694},{"id":"https://openalex.org/keywords/llava","display_name":"LLaVA","score":0.8299999833106995},{"id":"https://openalex.org/keywords/instructblip","display_name":"InstructBLIP","score":0.9419999718666077},{"id":"https://openalex.org/keywords/qwen2-5-vl","display_name":"Qwen2.5-VL","score":0.9319999814033508}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5778999924659729},{"id":"https://openalex.org/C2776207758","wikidata":"https://www.wikidata.org/wiki/Q5303302","display_name":"Downstream (manufacturing)","level":2,"score":0.42340001463890076},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.41179999709129333},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.39820000529289246},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3594000041484833},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3319000005722046},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.29660001397132874},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.29600000381469727},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.28439998626708984},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.2831999957561493},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.27730000019073486},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.27549999952316284}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.21203/rs.3.rs-11336583/v1","doi":"https://doi.org/10.21203/rs.3.rs-11336583/v1","is_oa":true,"landing_page_url":"https://doi.org/10.21203/rs.3.rs-11336583/v1","pdf_url":"https://www.researchsquare.com/article/rs-11336583/latest.pdf","source":{"id":"https://openalex.org/S4306525896","display_name":"Research Square","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":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Research Square","raw_type":"posted-content"}],"best_oa_location":{"id":"doi:10.21203/rs.3.rs-11336583/v1","doi":"https://doi.org/10.21203/rs.3.rs-11336583/v1","is_oa":true,"landing_page_url":"https://doi.org/10.21203/rs.3.rs-11336583/v1","pdf_url":"https://www.researchsquare.com/article/rs-11336583/latest.pdf","source":{"id":"https://openalex.org/S4306525896","display_name":"Research Square","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":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Research Square","raw_type":"posted-content"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7225299095.pdf","grobid_xml":"https://content.openalex.org/works/W7225299095.grobid-xml"},"referenced_works_count":12,"referenced_works":["https://openalex.org/W4366850747","https://openalex.org/W4385374425","https://openalex.org/W4386185600","https://openalex.org/W4390041933","https://openalex.org/W4402509982","https://openalex.org/W4404352832","https://openalex.org/W4404450583","https://openalex.org/W4407598818","https://openalex.org/W4415536336","https://openalex.org/W4417261316","https://openalex.org/W4417286665","https://openalex.org/W7111200874"],"related_works":["https://openalex.org/W4415316409","https://openalex.org/W4415536336","https://openalex.org/W7166854397","https://openalex.org/W4416830346","https://openalex.org/W4417286665","https://openalex.org/W6929485330","https://openalex.org/W4415975210","https://openalex.org/W6891885680","https://openalex.org/W7127078993","https://openalex.org/W4414854691"],"abstract_inverted_index":{"Abstract":[0],"Vision-language":[1],"models":[2,38],"(VLMs)":[3],"have":[4,22],"achieved":[5],"remarkable":[6],"success":[7],"across":[8],"a":[9,58,74,90,99],"wide":[10],"range":[11],"of":[12,142],"applications.However,":[13],"as":[14],"their":[15],"deployment":[16],"becomes":[17],"increasingly":[18],"widespread,":[19],"recent":[20],"studies":[21],"revealed":[23],"that":[24,79,141],"VLMs":[25],"are":[26],"vulnerable":[27],"to":[28,39,46,110,140],"visual":[29],"jailbreak":[30],"attacks,":[31],"where":[32],"adversarially":[33],"manipulated":[34],"images":[35],"induce":[36],"safety-aligned":[37],"produce":[40],"harmful":[41],"responses,":[42],"posing":[43],"serious":[44],"threats":[45],"real-world":[47],"security.Input":[48],"purification":[49,82],"defenses":[50],"must":[51],"balance":[52],"perturbation":[53],"suppression":[54],"against":[55],"content":[56],"preservation;":[57],"fixed":[59],"transformation":[60],"strength":[61,83],"may":[62],"be":[63],"unsuitable":[64],"for":[65],"inputs":[66],"and":[67,119],"attacks":[68,131],"with":[69],"different":[70],"characteristics.We":[71],"propose":[72],"FASDP,":[73],"Frequency-Aware":[75],"Scale-Diverse":[76],"Purification":[77],"framework":[78,103],"adaptively":[80],"allocates":[81],"per":[84],"image.It":[85],"fuses":[86],"multi-scale":[87],"candidates":[88],"from":[89],"single":[91],"frozen":[92],"denoiser":[93],"using":[94],"low-frequency":[95],"consistency":[96],"modulated":[97],"by":[98],"self-referenced":[100],"frequency":[101],"cue.The":[102],"requires":[104],"no":[105],"task-specific":[106],"training":[107],"or":[108],"access":[109],"the":[111,143],"target":[112],"model.We":[113],"evaluate":[114],"FASDP":[115],"on":[116,136],"LLaVA-1.5-7B,":[117],"InstructBLIP-7B,":[118],"Qwen2.5-VL-7B":[120],"under":[121,129],"three":[122],"representative":[123],"pixel-space":[124],"attacks.FASDP":[125],"attains":[126],"strong":[127],"protection":[128],"these":[130],"while":[132],"keeping":[133],"clean-image":[134],"performance":[135],"downstream":[137],"tasks":[138],"close":[139],"undefended":[144],"models.":[145]},"counts_by_year":[],"updated_date":"2026-11-09T13:22:03.289127","created_date":"2026-11-06T00:00:00"}