{"id":"W4382053191","doi":"10.1109/iwbf57495.2023.10157651","title":"On the influence of the quality of pseudo-labels on the self-supervised speaker verification task: a thorough analysis","year":2023,"lang":"en","type":"article","venue":"","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"ca_institutions":"Computer Research Institute of Montréal","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada","keywords":"Overfitting; Computer science; Discriminative model; Task (project management); Artificial intelligence; Noise (video); Memorization; Speech recognition; Quality (philosophy); Cluster analysis; Embedding; Pattern recognition (psychology); Machine learning; Artificial neural network; Mathematics; Image (mathematics); Engineering","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":[],"consensus_categories":[],"category_scores_codex":[0.001842926,0.0001143924,0.0001969804,0.0001080163,0.0001650203,0.00005654819,0.00131383,0.00004722315,0.0001808172],"category_scores_gemma":[0.0006499005,0.00004907613,0.0002335091,0.002881252,0.00009381678,0.0001147245,0.0001367097,0.0001144842,0.0001495862],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002113386,"about_ca_system_score_gemma":0.00004894525,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001724442,"about_ca_topic_score_gemma":0.00005550594,"domain_scores_codex":[0.9978691,0.0006687297,0.0003877871,0.0002786575,0.000641306,0.0001543706],"domain_scores_gemma":[0.9961416,0.001971381,0.0002233936,0.001455153,0.000181004,0.00002742667],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"observational","study_design_scores_codex":[0.00005688906,0.0006462153,0.008488889,0.00004744624,0.001074609,0.000001318974,0.006800189,0.002223608,0.04303994,0.8918062,0.004321062,0.04149364],"study_design_scores_gemma":[0.0002591663,0.0000750522,0.7840594,0.00004883927,0.0001517306,8.430703e-7,0.0008957845,0.1238436,0.07916097,0.01096184,0.0003020779,0.0002406624],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9814267,0.00000204915,0.002782194,0.009700015,0.00004942253,0.0002734101,0.00001091873,0.00009849171,0.005656858],"genre_scores_gemma":[0.996801,0.00001670225,0.001279474,0.001566334,0.000007553048,0.00002911902,0.000001264818,0.000004505725,0.0002940749],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8808444,"threshold_uncertainty_score":0.2441445,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05485249341111849,"score_gpt":0.294590206193768,"score_spread":0.2397377127826495,"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."}}