{"id":"W3136555065","doi":"10.1016/j.patrec.2021.03.009","title":"Perception matters: Exploring imperceptible and transferable anti-forensics for GAN-generated fake face imagery detection","year":2021,"lang":"en","type":"article","venue":"Pattern Recognition Letters","topic":"Digital Media Forensic Detection","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of British Columbia","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Artificial intelligence; Adversarial system; Perception; Face (sociological concept); Face detection; Deep learning; Benchmark (surveying); Computer vision; Facial recognition system; Machine learning; Pattern recognition (psychology); Psychology","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0002183949,0.0002701785,0.0002534742,0.0002188787,0.0002226626,0.0005835064,0.0001490813,0.0000889907,0.000034193],"category_scores_gemma":[0.00004364923,0.000305914,0.0001286431,0.0004007398,0.00006736638,0.00218658,0.00004719999,0.0001869619,0.00009682677],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001071454,"about_ca_system_score_gemma":0.00002364379,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003508376,"about_ca_topic_score_gemma":0.00003984912,"domain_scores_codex":[0.9981061,0.00009075977,0.0003561961,0.0007241023,0.0002752011,0.0004477041],"domain_scores_gemma":[0.9991571,0.0001071486,0.00008682316,0.0003070467,0.0002074281,0.0001344015],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.000007067992,0.00002356395,0.0000938593,0.00006872418,0.00002147526,0.00001527557,0.0006038109,0.00004909032,0.4335025,0.000001063976,0.0003438888,0.5652697],"study_design_scores_gemma":[0.00189935,0.0002396599,0.006371282,0.0002398275,0.0000784966,0.0004299522,0.001251102,0.02770867,0.9588143,0.0006198819,0.001332169,0.001015357],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.4938145,0.000009893711,0.502381,0.002343709,0.0009891751,0.0002218586,0.0000268224,0.0001871128,0.00002598017],"genre_scores_gemma":[0.9826657,0.00008753674,0.01007973,0.006385146,0.0002925771,0.0002522874,0.0001677826,0.00004522728,0.00002399463],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.5642543,"threshold_uncertainty_score":0.9999393,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04359750057162896,"score_gpt":0.2272622417478511,"score_spread":0.1836647411762222,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}