{"id":"W4401110548","doi":"10.1109/cai59869.2024.00186","title":"On the influence of metric learning loss functions for robust self-supervised speaker verification to label noise","year":2024,"lang":"en","type":"article","venue":"","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":0,"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":"Computer science; Metric (unit); Speech recognition; Noise (video); Speaker verification; Speaker recognition; Artificial intelligence; Robustness (evolution); Pattern recognition (psychology); 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":[0.0003828984,0.00007864652,0.00008057569,0.0002380235,0.0001190663,0.0001441367,0.0003106558,0.00003223472,0.0001239061],"category_scores_gemma":[0.0004083558,0.00005283534,0.00005770995,0.001275043,0.00001325757,0.0002050388,0.00004186861,0.00008358131,0.0004608082],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003233694,"about_ca_system_score_gemma":0.00004318804,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001774709,"about_ca_topic_score_gemma":0.000004881026,"domain_scores_codex":[0.9992206,0.00004950176,0.0001588226,0.0002570109,0.0001882482,0.0001258753],"domain_scores_gemma":[0.9986739,0.00081925,0.00002539485,0.0002793797,0.000152207,0.00004982377],"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":[0.00005086868,0.0004941736,0.0005826977,0.0001859857,0.0002099887,0.000006023266,0.002630481,0.02876104,0.01490592,0.2787858,0.01601314,0.6573739],"study_design_scores_gemma":[0.0002621131,0.0002671099,0.005264632,0.00009064208,0.00003915755,0.000008167654,0.0002188264,0.9616065,0.01069572,0.001147187,0.02014497,0.0002549392],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2605787,0.00002398524,0.7311752,0.003621026,0.000177392,0.0003394479,0.000004362842,0.0003081934,0.003771658],"genre_scores_gemma":[0.8945758,0.00001084437,0.1028092,0.0007495487,0.00002657584,0.00009873533,0.000002143084,0.000009813115,0.001717378],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9328455,"threshold_uncertainty_score":0.5922911,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03583955789924512,"score_gpt":0.2552282605859905,"score_spread":0.2193887026867454,"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."}}