{"id":"W3093096945","doi":"10.1145/3422852.3423486","title":"Intra and Inter-modality Interactions for Audio-visual Event Detection","year":2020,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Computer science; Modality (human–computer interaction); Audio visual; Modalities; Event (particle physics); Stimulus modality; Artificial intelligence; Visualization; Speech recognition; Audio signal processing; Audio signal; Sensory system; Speech coding; Multimedia","routes":{"ca_aff":true,"ca_fund":false,"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":[],"consensus_categories":[],"category_scores_codex":[0.00008843742,0.00006560191,0.0000791187,0.00002323588,0.0001155614,0.0001384375,0.0001286299,0.00001956816,0.00002022868],"category_scores_gemma":[0.00005020773,0.00005677574,0.00003370442,0.0001135995,0.00001791717,0.0004341969,0.0001201458,0.00007207047,0.000007854291],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001810218,"about_ca_system_score_gemma":0.00002032541,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001145755,"about_ca_topic_score_gemma":0.0000181688,"domain_scores_codex":[0.9994492,0.0000158019,0.0001296117,0.0002375805,0.00005897307,0.0001088164],"domain_scores_gemma":[0.9997131,0.0000460081,0.00004502614,0.00007493984,0.00004055519,0.00008031878],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00001488991,0.00003075389,0.0001070393,0.00005203721,0.0000138877,7.689405e-7,0.001170571,0.00001263384,0.01329514,0.003241279,0.001352169,0.9807088],"study_design_scores_gemma":[0.0005419358,0.0003207973,0.001281619,0.00003320307,0.00001316602,0.00002162858,0.000220336,0.8689802,0.08858859,0.00615519,0.03357186,0.0002714607],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02894092,0.00001347131,0.9565393,0.01356444,0.0002224388,0.00009602903,5.620111e-7,0.0001208521,0.0005019695],"genre_scores_gemma":[0.9805877,0.000001655561,0.01633223,0.00281645,0.0001359525,0.000014723,5.490779e-7,0.00000339544,0.000107352],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9804374,"threshold_uncertainty_score":0.2315248,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03034233145706322,"score_gpt":0.3061465161268468,"score_spread":0.2758041846697836,"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."}}