{"id":"https://openalex.org/W7163160714","doi":"https://doi.org/10.48550/arxiv.2606.01843","title":"Suppressing Forgery-Specific Shortcuts for Generalizable Deepfake Detection","display_name":"Suppressing Forgery-Specific Shortcuts for Generalizable Deepfake Detection","publication_year":2026,"publication_date":"2026-06-01","ids":{"openalex":"https://openalex.org/W7163160714","doi":"https://doi.org/10.48550/arxiv.2606.01843"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.01843","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.01843","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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.2606.01843","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137680028","display_name":"Yihui Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Yihui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137616262","display_name":"Yonghui Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Yonghui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137644553","display_name":"Jilong Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Jilong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137631322","display_name":"Fengbin Zhu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhu, Fengbin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137713416","display_name":"Le Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Le","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5137695957","display_name":"Tat-Seng Chua","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chua, Tat-Seng","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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.9440000057220459,"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"}},"topics":[{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.9440000057220459,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.022700000554323196,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.006000000052154064,"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/generalization","display_name":"Generalization","score":0.7452999949455261},{"id":"https://openalex.org/keywords/spurious-relationship","display_name":"Spurious relationship","score":0.7342000007629395},{"id":"https://openalex.org/keywords/subspace-topology","display_name":"Subspace topology","score":0.6715999841690063},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6101999878883362},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.4569000005722046},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.45570001006126404},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.4546000063419342}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7717000246047974},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.7452999949455261},{"id":"https://openalex.org/C97256817","wikidata":"https://www.wikidata.org/wiki/Q1462316","display_name":"Spurious relationship","level":2,"score":0.7342000007629395},{"id":"https://openalex.org/C32834561","wikidata":"https://www.wikidata.org/wiki/Q660730","display_name":"Subspace topology","level":2,"score":0.6715999841690063},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6101999878883362},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5552999973297119},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4997999966144562},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.4569000005722046},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.45570001006126404},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.4546000063419342},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.35350000858306885},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.35280001163482666},{"id":"https://openalex.org/C22789450","wikidata":"https://www.wikidata.org/wiki/Q420904","display_name":"Singular value decomposition","level":2,"score":0.35030001401901245},{"id":"https://openalex.org/C94124525","wikidata":"https://www.wikidata.org/wiki/Q912550","display_name":"Categorization","level":2,"score":0.3483999967575073},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.31299999356269836},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.29490000009536743},{"id":"https://openalex.org/C116834253","wikidata":"https://www.wikidata.org/wiki/Q2039217","display_name":"Identification (biology)","level":2,"score":0.2919999957084656},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.2653000056743622},{"id":"https://openalex.org/C2780148112","wikidata":"https://www.wikidata.org/wiki/Q1432581","display_name":"Proxy (statistics)","level":2,"score":0.2563999891281128}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.01843","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.01843","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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.2606.01843","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.01843","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":[{"display_name":"Reduced inequalities","score":0.5693879723548889,"id":"https://metadata.un.org/sdg/10"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Deepfake":[0],"detection":[1],"suffers":[2],"from":[3],"poor":[4],"generalization":[5,170,187],"across":[6],"forgery":[7,75,100],"methods,":[8],"as":[9,83],"existing":[10],"models":[11],"tend":[12],"to":[13,21,23,30,38,109,124,142],"rely":[14,143],"on":[15,116,144,177],"spurious":[16],"method-specific":[17,62,78,88],"shortcuts":[18,43,63],"that":[19,57,71,159,181],"fail":[20],"transfer":[22],"unseen":[24],"manipulations.":[25],"While":[26],"recent":[27],"approaches":[28],"attempt":[29],"improve":[31],"generalization,":[32],"they":[33],"lack":[34],"an":[35,84],"explicit":[36],"mechanism":[37],"identify":[39],"and":[40,60,80,103],"suppress":[41,132],"such":[42],"in":[44,136],"learned":[45],"representations.":[46],"In":[47],"this":[48,91,117],"work,":[49],"we":[50,93,119,130,154],"propose":[51],"Shortcut":[52],"Subspace":[53],"Suppression":[54],"(S^3)":[55],"framework":[56],"explicitly":[58],"characterizes":[59],"suppresses":[61],"via":[64],"subspace":[65,135],"modeling.":[66],"Our":[67],"key":[68],"insight":[69],"is":[70],"variations":[72],"distinguishing":[73],"different":[74],"methods":[76],"capture":[77],"artifacts":[79],"thus":[81],"serve":[82],"effective":[85],"proxy":[86],"for":[87,99,148],"shortcuts.":[89],"To":[90],"end,":[92],"train":[94],"a":[95,156],"lightweight":[96],"linear":[97],"probe":[98],"method":[101,183],"classification":[102],"perform":[104],"Singular":[105],"Value":[106],"Decomposition":[107],"(SVD)":[108],"extract":[110],"the":[111,133,140,164,201],"dominant":[112],"shortcut":[113,126,134,166],"subspace.":[114],"Building":[115],"formulation,":[118],"develop":[120],"two":[121],"complementary":[122],"strategies":[123],"reduce":[125],"reliance.":[127],"During":[128],"training,":[129],"softly":[131],"feature":[137],"representations,":[138],"encouraging":[139],"model":[141],"more":[145],"generalizable":[146],"cues":[147],"real/fake":[149],"discrimination.":[150],"At":[151],"inference":[152],"time,":[153],"introduce":[155],"training-free":[157],"counterpart":[158],"attenuates":[160],"neurons":[161],"aligned":[162],"with":[163,172],"identified":[165],"directions,":[167],"enabling":[168],"plug-and-play":[169],"enhancement":[171],"improved":[173],"interpretability.":[174],"Extensive":[175],"experiments":[176],"multiple":[178],"benchmarks":[179],"demonstrate":[180],"our":[182],"significantly":[184],"improves":[185],"cross-method":[186],"while":[188],"maintaining":[189],"strong":[190],"in-domain":[191],"performance.":[192],"The":[193],"code":[194],"will":[195],"be":[196],"released":[197],"upon":[198],"acceptance":[199],"of":[200],"submission.":[202]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-06-03T00:00:00"}