{"id":"W4402402651","doi":"10.1109/icasspw62465.2024.10669912","title":"Focal Modulation Networks for Interpretable Sound Classification","year":2024,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Concordia University","funders":"","keywords":"Modulation (music); Computer science; Sound (geography); Speech recognition; Artificial intelligence; Acoustics; Physics","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":[0.0001530426,0.00005006912,0.00004732017,0.00003691453,0.00007230254,0.0004640894,0.0001675131,0.00003594512,0.00001456335],"category_scores_gemma":[0.000008681539,0.00004087089,0.0000325582,0.0001641739,0.00001061801,0.000588463,0.00003762652,0.00004468301,0.00001580937],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002425208,"about_ca_system_score_gemma":0.00002642436,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000002027604,"about_ca_topic_score_gemma":0.000001936076,"domain_scores_codex":[0.9994925,0.000007319178,0.0001023938,0.0002155886,0.00006579631,0.0001163524],"domain_scores_gemma":[0.9997348,0.00005946093,0.00001674477,0.0001346527,0.00003024841,0.0000240732],"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":[0.000001732835,0.000007277014,0.00003106504,0.00003204948,0.000006920046,3.596111e-7,0.0001979685,0.0032251,0.0004006997,0.4408231,0.009597478,0.5456762],"study_design_scores_gemma":[0.0000324831,0.00001159924,0.0001736152,0.00002217171,0.000002283518,0.000001870634,0.000006520768,0.9526597,0.00008996378,0.03904786,0.007896729,0.00005525572],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.000612694,0.0001897759,0.9906566,0.001436831,0.0005802431,0.00008797347,2.153376e-7,0.0002584741,0.006177263],"genre_scores_gemma":[0.9489921,0.000002230428,0.04905722,0.0004247739,0.0001621797,0.0000198377,0.000003650378,0.000004735402,0.001333258],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9494345,"threshold_uncertainty_score":0.4475225,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03658115416600052,"score_gpt":0.2885343324847543,"score_spread":0.2519531783187538,"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."}}