{"id":"W2750726423","doi":"","title":"Natural Value Approximators: Learning when to Trust Past Estimates","year":2017,"lang":"en","type":"article","venue":"Neural Information Processing Systems","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Observability; Reinforcement learning; Bellman equation; Classification of discontinuities; Interpolation (computer graphics); Computer science; Value (mathematics); Artificial neural network; Discontinuity (linguistics); Function approximation; Inductive bias; Artificial intelligence; Function (biology); Mathematical optimization; Algorithm; Machine learning; Mathematics; Applied mathematics; Image (mathematics); Multi-task learning","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003959986,0.001022803,0.001329864,0.0006717373,0.0004371596,0.001283893,0.002375257,0.002274375,0.003649106],"category_scores_gemma":[0.02642677,0.0008120319,0.0005111183,0.000460117,0.001617641,0.00390638,0.00225737,0.003367678,0.0006732197],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008670982,"about_ca_system_score_gemma":0.001256969,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001892046,"about_ca_topic_score_gemma":0.002502193,"domain_scores_codex":[0.9987316,0.0004356338,0.0001117353,0.0003336972,0.0002595171,0.0001279285],"domain_scores_gemma":[0.990567,0.006690543,0.0007687417,0.0007959629,0.0008481665,0.000329506],"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":[0.0006327587,0.0002638164,0.003331764,0.0002714798,0.0001281692,0.0002239247,0.0004667949,0.6351255,0.005349016,0.08369867,0.004846069,0.2656622],"study_design_scores_gemma":[0.00002183285,0.00004590535,0.000083616,0.00002469674,0.000007173775,0.0000240032,0.00001541251,0.9840825,0.0009631889,0.01426285,0.0004590279,0.000009778083],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02005196,0.0002256433,0.9774762,0.0002979453,0.00005856693,0.00005837544,0.00003594158,0.0003699525,0.001425505],"genre_scores_gemma":[0.7244484,0.0002395761,0.2709734,0.0002780495,0.0001179071,0.0002940262,0.0001258327,0.0001783541,0.003344543],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003959986,"threshold_uncertainty_score":0.02094263,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01702164821844596,"score_gpt":0.264931605515689,"score_spread":0.247909957297243,"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."}}