{"id":"W4412534222","doi":"10.1088/2632-2153/adf278","title":"Feature learning and generalization in deep networks with orthogonal weights","year":2025,"lang":"en","type":"article","venue":"Machine Learning Science and Technology","topic":"Neural Networks and Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Vector Institute; University of Toronto","funders":"High Energy Physics","keywords":"Generalization; Feature (linguistics); Artificial intelligence; Pattern recognition (psychology); Feature learning; Computer science; Deep learning; Mathematics; Philosophy","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":[],"consensus_categories":[],"category_scores_codex":[0.0002790584,0.00009473133,0.0001083328,0.0004401719,0.0005424329,0.0001401825,0.0003208406,0.0000813721,9.559971e-7],"category_scores_gemma":[0.00004867706,0.00007276996,0.000005628372,0.003174402,0.0003598656,0.0002086265,0.0002828169,0.0005034031,4.738597e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001690336,"about_ca_system_score_gemma":0.00004477468,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001247898,"about_ca_topic_score_gemma":0.0000975623,"domain_scores_codex":[0.9990949,0.00002647328,0.00007991021,0.000427854,0.0001193641,0.0002515023],"domain_scores_gemma":[0.9996721,0.00003682126,0.00004970279,0.0001347296,0.00007245098,0.00003420518],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000003320552,0.00001381924,0.4097114,0.000005352993,0.000003176595,0.000008344156,0.0000560126,0.01455423,0.0007104365,0.3574542,0.00002074552,0.217459],"study_design_scores_gemma":[0.0001993701,0.00007378171,0.01858641,0.00002652125,0.000002641018,0.00003480139,0.00002010358,0.96763,0.000114754,0.003175709,0.01003196,0.0001039796],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5202516,0.001922672,0.4565191,0.01997322,0.00005898262,0.0001844925,9.748528e-8,0.0003399738,0.0007499204],"genre_scores_gemma":[0.9902195,0.0002004805,0.008998873,0.0001803432,0.00001076178,0.00001610874,0.000001468782,0.000003307985,0.0003691819],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9530757,"threshold_uncertainty_score":0.4172011,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.002308074299525024,"score_gpt":0.2194735595103882,"score_spread":0.2171654852108632,"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."}}