{"id":"W2980492205","doi":"10.1109/mlsp.2019.8918712","title":"Audio-Visual Fusion And Conditioning With Neural Networks For Event Recognition","year":2019,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Sherbrooke","funders":"","keywords":"Computer science; Modality (human–computer interaction); Pooling; Event (particle physics); Concatenation (mathematics); Speech recognition; Artificial intelligence; Modalities; Feature (linguistics); Audio visual; Audio signal processing; Pattern recognition (psychology); Audio signal; Multimedia; Speech coding","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":[0.0001065922,0.00006512238,0.00007470309,0.00002666942,0.0001210451,0.0001308687,0.00007220472,0.00002692983,0.000037463],"category_scores_gemma":[0.000004046992,0.00004828529,0.0000151741,0.00008286365,0.00001444681,0.0004125968,0.00004969904,0.00004701078,0.000005677838],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000007816229,"about_ca_system_score_gemma":0.00001385077,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000003383688,"about_ca_topic_score_gemma":0.000002786794,"domain_scores_codex":[0.9994764,0.00001127388,0.00008375285,0.0002092985,0.00008063459,0.0001386365],"domain_scores_gemma":[0.9997373,0.00004984633,0.00005388176,0.00007196132,0.00004948337,0.00003755877],"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":[0.00005935052,0.00005404913,0.003774944,0.00009517268,0.00001925595,0.000003946423,0.0004286006,0.004144747,0.00142044,0.003716127,0.002086436,0.9841969],"study_design_scores_gemma":[0.0005786017,0.0002366721,0.00178984,0.00006194307,0.000004882056,0.00002214006,0.00003578925,0.9951267,0.0008019988,0.0009313807,0.0002798165,0.0001302408],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2817308,0.00002846774,0.7169866,0.0003557722,0.0001108531,0.0001342144,2.981104e-7,0.00005196225,0.0006010357],"genre_scores_gemma":[0.9784677,0.000002693605,0.02004539,0.001085223,0.00006995902,0.00001152694,0.00001002065,0.000004546835,0.0003029154],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9909819,"threshold_uncertainty_score":0.1969017,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01382782289089173,"score_gpt":0.2481879598178739,"score_spread":0.2343601369269822,"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."}}