{"id":"W4407241185","doi":"10.1007/s10994-024-06670-4","title":"Schema-tune: noise-driven bias mitigation in transformer-based language models","year":2025,"lang":"en","type":"article","venue":"Machine Learning","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"Toronto Metropolitan University","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Schema (genetic algorithms); Transformer; Artificial intelligence; Natural language processing; Machine learning; Speech recognition; Engineering; Electrical 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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002396497,0.0007791496,0.0008380673,0.0006615054,0.000366512,0.001402978,0.001718941,0.0009971495,0.004400307],"category_scores_gemma":[0.01063714,0.0005016929,0.001002277,0.0007644337,0.0004851278,0.002673563,0.001937225,0.002135118,0.00303412],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005232056,"about_ca_system_score_gemma":0.001284725,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002324363,"about_ca_topic_score_gemma":0.005052336,"domain_scores_codex":[0.9988698,0.0004668337,0.00007010066,0.0002887622,0.0002107476,0.00009364881],"domain_scores_gemma":[0.997243,0.001476369,0.00009693949,0.0006679251,0.0004128926,0.0001028516],"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":[0.001206043,0.0003091511,0.00436962,0.0005877334,0.0004370085,0.0002750089,0.0007349729,0.1623342,0.04055192,0.05476939,0.03343135,0.7009937],"study_design_scores_gemma":[0.00006303724,0.000056581,0.0002476902,0.00001953259,0.00005996212,0.00009702452,0.0000648436,0.9461794,0.01331811,0.03537262,0.004501036,0.00002020706],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01330603,0.0002733179,0.9765866,0.0002403531,0.0001012959,0.00006321753,0.0006174067,0.007772401,0.00103946],"genre_scores_gemma":[0.5386167,0.000510439,0.4492496,0.0005793054,0.0001643159,0.0002379145,0.003793854,0.002880503,0.003967382],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004400307,"threshold_uncertainty_score":0.0147205,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01802481198386612,"score_gpt":0.2643060661716281,"score_spread":0.246281254187762,"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."}}