{"id":"W4380366610","doi":"10.48550/arxiv.2306.05991","title":"Approximate information state based convergence analysis of recurrent Q-learning","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Memory and Neural Computing","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Alliance de recherche numérique du Canada","keywords":"Convergence (economics); Reinforcement learning; Computer science; Markov decision process; Representation (politics); Observable; Partially observable Markov decision process; Recurrent neural network; Markov chain; Markov process; Artificial intelligence; Machine learning; Mathematical optimization; Algorithm; Markov model; Artificial neural network; Mathematics; Statistics","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":[0.0001546435,0.0001917182,0.0003319285,0.0005828319,0.00005767974,0.00001564033,0.0002500835,0.00009453888,0.00002540838],"category_scores_gemma":[0.00003613773,0.0002398652,0.0002001003,0.001196741,0.0000327445,0.0002233902,0.0002085191,0.0004636156,0.00002500307],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009436128,"about_ca_system_score_gemma":0.00002446707,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001406846,"about_ca_topic_score_gemma":0.000009713828,"domain_scores_codex":[0.9991859,0.00004744849,0.0002576961,0.0002426415,0.00006403127,0.0002023171],"domain_scores_gemma":[0.9992573,0.00009671629,0.0002034226,0.0002866525,0.00009036847,0.0000655684],"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":[0.00002166388,0.000005259197,0.001566157,0.0003127182,0.0002557175,0.0000123349,0.000149054,0.9968373,0.0001028029,0.0001984394,0.00000961211,0.0005289438],"study_design_scores_gemma":[0.0001346866,0.00001592734,0.001536205,0.00008268427,0.000317645,8.358493e-8,0.00007231439,0.9954756,0.001588421,0.0004801009,0.00007502914,0.0002213325],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6019627,0.00001181385,0.397116,0.000001820495,0.0002865867,0.00009804397,0.00003296324,0.000355374,0.0001347268],"genre_scores_gemma":[0.9993663,0.0001660954,0.0001932287,0.000005291291,0.000009224696,5.657183e-7,0.0001589684,0.00001519089,0.00008517562],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3974036,"threshold_uncertainty_score":0.9781419,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06678416009321476,"score_gpt":0.1896028334901685,"score_spread":0.1228186733969537,"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."}}