{"id":"W2804948070","doi":"","title":"Using Reward Machines for High-Level Task Specification and Decomposition in Reinforcement Learning","year":2018,"lang":"en","type":"article","venue":"International Conference on Machine Learning","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":129,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Reinforcement learning; Computer science; Task (project management); Decomposition; Artificial intelligence; Human–computer interaction; Engineering","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":[0.0005974613,0.000226202,0.0002000981,0.0004187428,0.0003053882,0.0003731153,0.0005681012,0.00008287372,0.00008011689],"category_scores_gemma":[0.000330457,0.0002333267,0.0000381215,0.000201728,0.00007441393,0.0005368039,0.0002301561,0.0004673228,0.00003255062],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000188603,"about_ca_system_score_gemma":0.00005438289,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002162561,"about_ca_topic_score_gemma":0.00003057971,"domain_scores_codex":[0.9981607,0.0001355366,0.0004595779,0.0005091034,0.0004370407,0.0002979974],"domain_scores_gemma":[0.9988728,0.0001722133,0.0003412985,0.0002151773,0.0003283149,0.00007024964],"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":[0.0001252948,0.00002815578,0.01054913,0.0000191087,0.00003666681,0.000004320693,0.0008235468,0.6796286,0.01381133,0.2791476,0.00002038995,0.01580594],"study_design_scores_gemma":[0.0008115113,0.0004897515,0.005755191,0.0001301114,0.000006088211,0.00001191976,0.00003789358,0.9876899,0.0006453297,0.003197852,0.0009762649,0.0002482276],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03878467,0.00001254822,0.9538448,0.001216866,0.000495991,0.0002713375,0.000002157976,0.0001187153,0.005252841],"genre_scores_gemma":[0.9404784,0.00003459271,0.05795851,0.0001652802,0.0002059745,0.00002163337,0.00009358234,0.00002031875,0.001021693],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9016938,"threshold_uncertainty_score":0.9514787,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.09333430035172953,"score_gpt":0.3548435333796292,"score_spread":0.2615092330278997,"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."}}