{"id":"W4379523790","doi":"10.2118/213104-ms","title":"Reinforcement Learning for Multi-Well SAGD Optimization: A Policy Gradient Approach","year":2023,"lang":"en","type":"article","venue":"","topic":"Reservoir Engineering and Simulation Methods","field":"Engineering","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Alberta","funders":"","keywords":"Reinforcement learning; Markov decision process; Computer science; Mathematical optimization; Process (computing); Convergence (economics); Artificial neural network; Markov process; Action (physics); Monte Carlo method; Bellman equation; Artificial intelligence; Mathematics","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002645097,0.0001259063,0.000127889,0.0002432644,0.00007625,0.00003751985,0.00009246323,0.00006036193,0.0000406176],"category_scores_gemma":[0.00009902301,0.0001211764,0.00006918854,0.0005112909,0.000007207402,0.00007252045,0.0000231091,0.00009249312,0.00003363491],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006281661,"about_ca_system_score_gemma":0.00001015684,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001069051,"about_ca_topic_score_gemma":3.233037e-7,"domain_scores_codex":[0.9992548,0.00001548975,0.0001991495,0.0001350934,0.0001111516,0.0002843649],"domain_scores_gemma":[0.9996524,0.00006643272,0.00001410475,0.0001496446,0.00003545161,0.00008191232],"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":[0.00000264973,0.000004541342,0.00003912392,0.00009176799,0.00002143755,2.856574e-7,0.0001926764,0.9973688,0.00004693648,0.001039393,0.0006996326,0.0004927361],"study_design_scores_gemma":[0.0005713819,0.00001943185,0.00007015749,0.00000633228,0.000004494627,7.320821e-7,0.00007056213,0.9857358,0.0002065103,0.00001246043,0.01315652,0.0001456029],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.001676062,0.00002653198,0.9861161,0.00003997347,0.0001407,0.0002709255,6.207659e-7,0.001429231,0.01029982],"genre_scores_gemma":[0.2397902,0.0001106818,0.7397505,0.00002559486,0.0002260733,0.000194849,0.0001338232,0.00007986841,0.01968843],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.2463656,"threshold_uncertainty_score":0.494143,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05155217862445684,"score_gpt":0.3101497248625225,"score_spread":0.2585975462380657,"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."}}