{"id":"W4416978430","doi":"10.1038/s41467-025-66983-3","title":"Sufficient is better than optimal for training neural networks","year":2025,"lang":"en","type":"article","venue":"Nature Communications","topic":"Stochastic Gradient Optimization Techniques","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Queen's University","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada; Universities Space Research Association","keywords":"Overfitting; Leverage (statistics); Spurious relationship; Artificial neural network; Deep neural networks; Convolutional neural network; Feedforward neural network; Training (meteorology)","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":[0.000233842,0.0001224477,0.0001314341,0.0001848993,0.0004105929,0.0001441187,0.002819721,0.0001928442,0.000002925025],"category_scores_gemma":[0.0001318118,0.0001224376,0.00009610711,0.00070507,0.00008880431,0.000175164,0.0006604489,0.0005278496,0.000001017095],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000464748,"about_ca_system_score_gemma":0.00005352204,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00000257776,"about_ca_topic_score_gemma":0.000008552351,"domain_scores_codex":[0.9991058,0.00005476555,0.0002245292,0.0002639481,0.0001159182,0.0002350119],"domain_scores_gemma":[0.9971921,0.0004694081,0.00008382311,0.001994202,0.0002164625,0.00004402855],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000005530454,0.0001231933,0.0004468079,0.000006599834,0.00004314616,2.637343e-7,0.00193867,0.01082636,0.00003094446,0.9315225,0.02645391,0.02860202],"study_design_scores_gemma":[0.0001590421,0.00002839727,0.000722799,0.00002068182,0.00001259919,0.000002095062,0.00005098611,0.9851558,0.0001144008,0.001994429,0.01162203,0.0001167578],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0003647555,0.001380255,0.9686041,0.02696795,0.0003849503,0.0004242884,0.000008632097,0.0003687907,0.001496215],"genre_scores_gemma":[0.5867033,0.00002247917,0.4094767,0.003504011,0.00001917242,0.0001453078,0.00002405727,0.000006350664,0.00009858925],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9743294,"threshold_uncertainty_score":0.523979,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03103649933153931,"score_gpt":0.3159515656064312,"score_spread":0.2849150662748919,"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."}}