{"id":"W3207572704","doi":"10.1109/icassp43922.2022.9746260","title":"Continual Learning Using Lattice-Free MMI for Speech Recognition","year":2022,"lang":"en","type":"article","venue":"ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"NeuroRx Research (Canada)","funders":"","keywords":"Forgetting; Computer science; Regularization (linguistics); Artificial neural network; Exploit; Speech recognition; Word error rate; Artificial intelligence; Acoustic model; Domain adaptation; Machine learning; Speech processing","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":["metaepi_narrow","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001448564,0.000466085,0.0005003641,0.0006060936,0.001287143,0.001068986,0.00157208,0.0001632202,0.00225299],"category_scores_gemma":[0.0006666324,0.0005147185,0.0002045393,0.0005093396,0.0001666012,0.0007854298,0.000597404,0.001118538,0.00004596077],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003654388,"about_ca_system_score_gemma":0.000428516,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004558901,"about_ca_topic_score_gemma":0.00001043349,"domain_scores_codex":[0.9954802,0.0003063848,0.0007829735,0.001159239,0.001600193,0.0006710199],"domain_scores_gemma":[0.9971681,0.0006217792,0.0006196291,0.0003975144,0.0009143074,0.0002786534],"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":[0.0004455429,0.0005357982,0.0001471585,0.0001081299,0.0001742031,0.000341671,0.0005208462,0.0005214434,0.07491359,0.002428193,0.004662226,0.9152012],"study_design_scores_gemma":[0.001997865,0.0007385536,0.00008095241,0.0001816527,0.0001113686,0.0005976132,0.001809882,0.9525791,0.01001969,0.024838,0.006090429,0.0009548264],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1944957,0.0003247609,0.7755654,0.006320996,0.004091332,0.00147259,0.001215051,0.0008423034,0.01567178],"genre_scores_gemma":[0.8670032,0.0001121499,0.1269774,0.001498994,0.000798629,0.0002231357,0.0002086158,0.0000846725,0.003093228],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9520577,"threshold_uncertainty_score":0.999968,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.09276504634484213,"score_gpt":0.3157352333647936,"score_spread":0.2229701870199515,"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."}}