{"id":"W2894812324","doi":"10.1016/j.disopt.2023.100795","title":"Principled deep neural network training through linear programming","year":2023,"lang":"en","type":"article","venue":"Discrete Optimization","topic":"Stochastic Gradient Optimization Techniques","field":"Computer Science","cited_by":25,"is_retracted":false,"has_abstract":false,"ca_institutions":"","funders":"Office of Naval Research; Institut de Valorisation des Données; National Science Foundation","keywords":"Computer science; Deep learning; Polyhedron; Artificial intelligence; Artificial neural network; Linear programming; Perspective (graphical); Dependency (UML); Task (project management); Representation (politics); Sample (material); Function (biology); Machine learning; Mathematical optimization; Algorithm; Theoretical computer science; Mathematics; Geometry","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003378717,0.0002128354,0.0002051654,0.0001379996,0.0003138495,0.0002227489,0.000615901,0.00009880913,0.00001916953],"category_scores_gemma":[0.0001722186,0.0002119005,0.00008418348,0.002416539,0.00005187233,0.000908109,0.0002759902,0.0001361203,0.00002416529],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004414388,"about_ca_system_score_gemma":0.00003597606,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000004824495,"about_ca_topic_score_gemma":0.000001781258,"domain_scores_codex":[0.998116,0.00007808072,0.0003756795,0.00051254,0.0003568407,0.0005609294],"domain_scores_gemma":[0.9990093,0.00008996257,0.0001831591,0.0004910667,0.000134621,0.00009184995],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000003862131,0.000008088374,0.00008195752,0.000007055783,0.00001048897,0.000005998399,0.001612352,0.955761,0.000003943111,0.03175353,0.00008937803,0.01066234],"study_design_scores_gemma":[0.0002319761,0.00007015122,0.00007238414,0.00002524846,0.00000911098,0.000007590823,0.0000997778,0.9972309,0.00003805693,0.001479221,0.0004936107,0.0002419125],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.0001871664,0.00004300162,0.9942153,0.0005607739,0.0004408024,0.0005386477,0.000001395622,0.002790446,0.001222432],"genre_scores_gemma":[0.0510679,0.00003382201,0.9480601,0.0001942266,0.0001787391,0.0001478244,0.0001580378,0.0000407762,0.0001185773],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.05088073,"threshold_uncertainty_score":0.8641052,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04017734624734458,"score_gpt":0.2895398856964629,"score_spread":0.2493625394491183,"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."}}