{"id":"W4387760932","doi":"10.1109/waspaa58266.2023.10248180","title":"Unsupervised Improvement of Audio-Text Cross-Modal Representations","year":2023,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Concordia University; Mila - Quebec Artificial Intelligence Institute","funders":"","keywords":"Computer science; Modal; Natural language processing; Artificial intelligence; Event (particle physics); Domain (mathematical analysis); Conjunction (astronomy); Speech recognition","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001910755,0.001751098,0.001098113,0.001999181,0.0006877955,0.001396907,0.002146726,0.001836829,0.003282799],"category_scores_gemma":[0.008042788,0.0004241707,0.001189281,0.001422479,0.000915189,0.003255106,0.003494554,0.003130374,0.002855812],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007158447,"about_ca_system_score_gemma":0.0009449112,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002633924,"about_ca_topic_score_gemma":0.004622549,"domain_scores_codex":[0.9983408,0.0004018121,0.00007186335,0.0007165982,0.0002705345,0.0001983516],"domain_scores_gemma":[0.996471,0.001374723,0.000329025,0.0008155307,0.0008111177,0.000198587],"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":[0.001182217,0.00108723,0.003638424,0.0003481064,0.000204756,0.0003811485,0.0005013625,0.08146774,0.078398,0.007087854,0.01580982,0.8098933],"study_design_scores_gemma":[0.00004478893,0.000199237,0.002012576,0.00003228004,0.0000633235,0.0001851061,0.0001961217,0.9529812,0.03232586,0.008020249,0.003897416,0.00004186529],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1005443,0.0009085491,0.8829284,0.0003910999,0.0002270612,0.0001092146,0.001015165,0.01010064,0.003775508],"genre_scores_gemma":[0.7298636,0.0005299663,0.246684,0.0006459384,0.0002707761,0.0002854957,0.009988965,0.001077695,0.0106536],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003282799,"threshold_uncertainty_score":0.01098204,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02687686276970769,"score_gpt":0.3153915764838867,"score_spread":0.288514713714179,"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."}}