{"id":"W4289792473","doi":"10.1109/jstsp.2022.3196562","title":"L-Mix: A Latent-Level Instance Mixup Regularization for Robust Self-Supervised Speaker Representation Learning","year":2022,"lang":"en","type":"article","venue":"IEEE Journal of Selected Topics in Signal Processing","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":22,"is_retracted":false,"has_abstract":true,"ca_institutions":"Computer Research Institute of Montréal","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Feature learning; Embedding; Speech recognition; Artificial intelligence; Regularization (linguistics); Speaker recognition; Pattern recognition (psychology); Supervised learning; Semi-supervised learning; Machine learning; Artificial neural network","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":[0.0007147576,0.0001202183,0.000225606,0.0003607397,0.0003487583,0.0001910813,0.0004187064,0.00005722836,0.00002842797],"category_scores_gemma":[0.0001620995,0.0001248899,0.00007442977,0.001297134,0.00001478573,0.0007969969,0.00004376631,0.0004631566,8.216085e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002134766,"about_ca_system_score_gemma":0.000330151,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00000259287,"about_ca_topic_score_gemma":0.000004023483,"domain_scores_codex":[0.9981977,0.0002321828,0.0005846494,0.0002368398,0.0005167488,0.0002318392],"domain_scores_gemma":[0.9984915,0.000128188,0.0005292705,0.00009277465,0.0006989633,0.00005925638],"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":[0.0002647232,0.0005601365,0.005761649,0.0001846348,0.0000919447,0.0001374726,0.006476828,0.07196908,0.03680374,0.0006096379,0.0004747392,0.8766654],"study_design_scores_gemma":[0.001720406,0.000300949,0.00250355,0.0001546028,0.00003146653,0.0002846938,0.0003632165,0.9722487,0.01724854,0.003904882,0.0009616309,0.0002773414],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0767728,0.0001775081,0.9215386,0.0007976821,0.0002670527,0.0001844456,0.000001529494,0.00005905598,0.000201354],"genre_scores_gemma":[0.6282171,0.00002801585,0.3707991,0.0002339626,0.0002956159,0.00002076199,0.000006736334,0.00002012929,0.0003785755],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9002796,"threshold_uncertainty_score":0.5092862,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05517068553173211,"score_gpt":0.2693890251686757,"score_spread":0.2142183396369436,"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."}}