{"id":"W4388904294","doi":"10.1016/j.ifacol.2023.10.923","title":"A modular framework for stabilizing deep reinforcement learning control","year":2023,"lang":"en","type":"article","venue":"IFAC-PapersOnLine","topic":"Model Reduction and Neural Networks","field":"Physics and Astronomy","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Honeywell (Canada); University of British Columbia","funders":"","keywords":"Reinforcement learning; Modular design; Computer science; Realization (probability); Parameterized complexity; Construct (python library); Stability (learning theory); Set (abstract data type); Artificial intelligence; Artificial neural network; Domain (mathematical analysis); Deep learning; Nonlinear system; Control (management); Control engineering; Machine learning; Engineering; Algorithm; Mathematics","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":[0.0001745112,0.0001855328,0.0002358514,0.00006209964,0.0002645628,0.00004852562,0.0001184625,0.00006754912,0.000576266],"category_scores_gemma":[0.00003342735,0.0001716968,0.0001936531,0.0002299879,0.00002959529,0.00009375947,0.00003340494,0.0003043052,0.00008192637],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002330369,"about_ca_system_score_gemma":0.00002226132,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0000200445,"about_ca_topic_score_gemma":0.000001313268,"domain_scores_codex":[0.9987988,0.00003462763,0.000252457,0.0003044117,0.000172379,0.0004373501],"domain_scores_gemma":[0.9993427,0.0001880406,0.000101191,0.0001880428,0.00006082545,0.0001191315],"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":[0.000118339,0.0000574806,0.001750463,0.00003092749,0.0001561299,0.000002155794,0.0006627776,0.9171149,0.004809348,0.02814718,0.00003641751,0.04711393],"study_design_scores_gemma":[0.001059331,0.0001071728,0.000152288,0.00004164907,0.00003927213,5.016145e-7,0.001074253,0.9879317,0.0003875069,0.003957113,0.005002785,0.0002463981],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1394466,0.00006999858,0.8568382,0.001496275,0.0004573444,0.0005815393,0.00001542901,0.0002379284,0.0008566955],"genre_scores_gemma":[0.9624369,0.00001487933,0.03393158,0.0002261412,0.001187275,0.0001410355,0.0001802049,0.00003928089,0.001842665],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8229903,"threshold_uncertainty_score":0.7001593,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02128174126494181,"score_gpt":0.2848373929165303,"score_spread":0.2635556516515885,"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."}}