{"id":"W4406284830","doi":"10.1038/s43246-024-00731-w","title":"Probing out-of-distribution generalization in machine learning for materials","year":2025,"lang":"en","type":"article","venue":"Communications Materials","topic":"Machine Learning in Materials Science","field":"Materials Science","cited_by":43,"is_retracted":false,"has_abstract":true,"ca_institutions":"Schwartz/Reisman Emergency Medicine Institute; Vector Institute; Structural Genomics Consortium; University of Toronto","funders":"Canada First Research Excellence Fund; University of Toronto; Alliance de recherche numérique du Canada; Western Canada Research Grid","keywords":"Generalization; Distribution (mathematics); Computer science; Artificial intelligence; Mathematics; Mathematical analysis","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":[0.003467127,0.0002091777,0.0004944274,0.0002114057,0.0003620225,0.0002647962,0.001314715,0.0001257753,0.0005330595],"category_scores_gemma":[0.001333758,0.0002092755,0.00004191641,0.000371129,0.0002532714,0.0003151006,0.0006002126,0.00008799597,0.00003631593],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001246163,"about_ca_system_score_gemma":0.00010779,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003880706,"about_ca_topic_score_gemma":0.000100494,"domain_scores_codex":[0.9970295,0.001011038,0.0010721,0.0003700543,0.000168774,0.0003485071],"domain_scores_gemma":[0.9976704,0.0003548459,0.0005472346,0.001164547,0.000229751,0.00003324821],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.00005716257,0.00008213974,0.0004402946,0.0002066719,0.000004670569,1.51146e-7,0.0002729861,0.001322726,0.9761139,0.02116678,0.0001784808,0.0001540192],"study_design_scores_gemma":[0.0005518013,0.00004990654,0.00125823,0.0002579733,0.00002316269,9.911951e-7,0.00005215441,0.002630898,0.9836841,0.002370569,0.008914511,0.0002056632],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9496672,0.0002952255,0.04382433,0.001135136,0.002552211,0.001353085,0.0004763546,0.0002140732,0.0004823372],"genre_scores_gemma":[0.9778547,0.0001251574,0.01988277,0.00006694339,0.00006221856,0.0004389761,0.001172597,0.00002365674,0.000372956],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.02818748,"threshold_uncertainty_score":0.8534008,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03469835939800099,"score_gpt":0.3325907825709555,"score_spread":0.2978924231729546,"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."}}