{"id":"https://openalex.org/W7166832117","doi":"https://doi.org/10.48550/arxiv.2606.30791","title":"Probing-Guided Layer Selection from Self-Supervised Speech Models for Generalizable Audio Deepfake Detection","display_name":"Probing-Guided Layer Selection from Self-Supervised Speech Models for Generalizable Audio Deepfake Detection","publication_year":2026,"publication_date":"2026-06-29","ids":{"openalex":"https://openalex.org/W7166832117","doi":"https://doi.org/10.48550/arxiv.2606.30791"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.30791","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.30791","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","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.2606.30791","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139826375","display_name":"Marjan Beheshti","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Beheshti, Marjan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5038105873","display_name":"M. Rostami","orcid":"https://orcid.org/0000-0002-2157-9539"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rostami, Majid","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139806947","display_name":"Bo Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Bo","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":0.25220000743865967,"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":0.25220000743865967,"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/T12357","display_name":"Digital Media Forensic Detection","score":0.22179999947547913,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.19449999928474426,"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/pooling","display_name":"Pooling","score":0.6453999876976013},{"id":"https://openalex.org/keywords/discriminative-model","display_name":"Discriminative model","score":0.6431000232696533},{"id":"https://openalex.org/keywords/layer","display_name":"Layer (electronics)","score":0.6292999982833862},{"id":"https://openalex.org/keywords/bottleneck","display_name":"Bottleneck","score":0.5411999821662903},{"id":"https://openalex.org/keywords/fuse","display_name":"Fuse (electrical)","score":0.5289999842643738},{"id":"https://openalex.org/keywords/classifier","display_name":"Classifier (UML)","score":0.5144000053405762},{"id":"https://openalex.org/keywords/selection","display_name":"Selection (genetic algorithm)","score":0.46000000834465027},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.3919999897480011}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7297000288963318},{"id":"https://openalex.org/C70437156","wikidata":"https://www.wikidata.org/wiki/Q7228652","display_name":"Pooling","level":2,"score":0.6453999876976013},{"id":"https://openalex.org/C97931131","wikidata":"https://www.wikidata.org/wiki/Q5282087","display_name":"Discriminative model","level":2,"score":0.6431000232696533},{"id":"https://openalex.org/C2779227376","wikidata":"https://www.wikidata.org/wiki/Q6505497","display_name":"Layer (electronics)","level":2,"score":0.6292999982833862},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.5411999821662903},{"id":"https://openalex.org/C141353440","wikidata":"https://www.wikidata.org/wiki/Q182221","display_name":"Fuse (electrical)","level":2,"score":0.5289999842643738},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5232999920845032},{"id":"https://openalex.org/C95623464","wikidata":"https://www.wikidata.org/wiki/Q1096149","display_name":"Classifier (UML)","level":2,"score":0.5144000053405762},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.5110999941825867},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.46000000834465027},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3919999897480011},{"id":"https://openalex.org/C40969351","wikidata":"https://www.wikidata.org/wiki/Q3516228","display_name":"Word error rate","level":2,"score":0.3463999927043915},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.34360000491142273},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3375000059604645},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.32749998569488525},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3248000144958496},{"id":"https://openalex.org/C148483581","wikidata":"https://www.wikidata.org/wiki/Q446488","display_name":"Feature selection","level":2,"score":0.32359999418258667},{"id":"https://openalex.org/C2984842247","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep neural networks","level":3,"score":0.29820001125335693},{"id":"https://openalex.org/C2779714256","wikidata":"https://www.wikidata.org/wiki/Q25305062","display_name":"Multiple Models","level":2,"score":0.26969999074935913},{"id":"https://openalex.org/C194051981","wikidata":"https://www.wikidata.org/wiki/Q1337691","display_name":"Economic shortage","level":3,"score":0.2639999985694885},{"id":"https://openalex.org/C93959086","wikidata":"https://www.wikidata.org/wiki/Q6888345","display_name":"Model selection","level":2,"score":0.257999986410141},{"id":"https://openalex.org/C28034677","wikidata":"https://www.wikidata.org/wiki/Q17092530","display_name":"Interleaving","level":2,"score":0.25110000371932983}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.30791","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.30791","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.30791","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.30791","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/10","score":0.7199029326438904,"display_name":"Reduced inequalities"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Audio":[0],"deepfake":[1],"detection":[2],"systems":[3],"often":[4],"fail":[5],"to":[6,16,147],"generalize":[7],"across":[8,112],"domains":[9],"because":[10],"they":[11],"rely":[12],"on":[13,177],"features":[14],"tied":[15],"specific":[17],"attacks":[18],"or":[19,37],"recording":[20],"conditions.":[21],"Self-supervised":[22],"speech":[23],"models":[24],"offer":[25],"rich":[26],"multi-layer":[27],"representations,":[28],"yet":[29],"existing":[30],"approaches":[31],"either":[32],"use":[33],"a":[34,51,81,88,102,157,187],"single":[35],"layer":[36,45,82,159],"fuse":[38],"all":[39,202],"layers":[40,96,123,165,204],"indiscriminately,":[41],"and":[42,101,179,198],"only":[43,93],"reveal":[44],"importance":[46],"after":[47],"training.":[48],"We":[49],"propose":[50],"model-agnostic,":[52],"two-stage":[53],"methodology":[54],"that":[55],"identifies":[56],"informative":[57,122],"depth":[58,126],"zones":[59,127],"before":[60],"any":[61],"task-specific":[62],"model":[63],"is":[64],"trained.":[65],"In":[66,84],"the":[67,85,94,107,115,150,192],"first":[68],"stage,":[69,87],"lightweight":[70],"XGBoost":[71],"probes":[72],"evaluate":[73],"each":[74],"transformer":[75],"layer's":[76],"cross-domain":[77,181],"discriminative":[78],"power,":[79],"producing":[80],"ranking.":[83],"second":[86],"compact":[89],"neural":[90],"classifier":[91],"fuses":[92],"selected":[95],"through":[97],"per-layer":[98],"attention":[99],"pooling":[100],"shared":[103,185],"bottleneck":[104],"projection,":[105],"while":[106,140],"backbone":[108],"remains":[109],"frozen.":[110],"Applied":[111],"three":[113],"backbones,":[114],"probing":[116,151],"reveals":[117],"two":[118],"key":[119],"findings.":[120],"First,":[121],"cluster":[124],"in":[125],"rather":[128,155],"than":[129,156],"at":[130],"uniquely":[131],"optimal":[132],"positions:":[133],"within-zone":[134],"substitutions":[135],"fall":[136],"within":[137],"multi-seed":[138],"noise,":[139],"zone":[141],"violations":[142],"degrade":[143],"performance":[144],"by":[145],"up":[146],"5x.":[148],"Second,":[149],"produces":[152],"backbone-specific":[153],"selections":[154],"fixed":[158],"recipe.":[160],"On":[161],"XLS-R-300M,":[162],"four":[163,184],"probing-selected":[164],"with":[166,205],"1.34M":[167],"trainable":[168],"parameters":[169],"achieve":[170],"4.94":[171],"+/-":[172],"0.32%":[173],"equal":[174],"error":[175],"rate":[176],"In-The-Wild":[178],"5.07%":[180],"average":[182],"over":[183,191],"datasets,":[186],"28%":[188],"relative":[189],"improvement":[190],"best":[193],"prior":[194],"frozen-backbone":[195],"result":[196],"(Xiao":[197],"Vu,":[199],"2025)":[200],"using":[201],"25":[203],"identical":[206],"training":[207],"data.":[208]},"counts_by_year":[],"updated_date":"2026-07-02T06:18:51.028212","created_date":"2026-07-02T00:00:00"}