{"id":"W4415903164","doi":"10.24963/kr.2025/55","title":"Pushdown Reward Machines for Reinforcement Learning","year":2025,"lang":"en","type":"article","venue":"","topic":"Machine Learning and Algorithms","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Vector Institute; University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada; Canadian Institute for Advanced Research","keywords":"Reinforcement learning; Exploit; Constant (computer programming); Pushdown automaton; Finite-state machine; Extension (predicate logic); Counterfactual thinking; Task (project management); Automaton","routes":{"ca_aff":true,"ca_fund":true,"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":[0.002327599,0.001127037,0.001006028,0.0006452682,0.0006742724,0.001729726,0.001613107,0.001586847,0.004578013],"category_scores_gemma":[0.01369083,0.0006330651,0.001181191,0.000638756,0.002206644,0.002698564,0.001688506,0.00366915,0.0008404184],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001548313,"about_ca_system_score_gemma":0.001332156,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001539728,"about_ca_topic_score_gemma":0.001799964,"domain_scores_codex":[0.9978816,0.0009090424,0.0001673098,0.0004920737,0.0003965327,0.0001535418],"domain_scores_gemma":[0.9905854,0.007408507,0.0004717374,0.0008444182,0.0004815436,0.0002084173],"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":[0.0002102329,0.00008613692,0.0008791531,0.0002615807,0.00008147599,0.0001481286,0.0001783116,0.6325661,0.003557902,0.294005,0.001720955,0.06630501],"study_design_scores_gemma":[0.00002057938,0.00005105167,0.00007476685,0.000019884,0.00001345594,0.00002841139,0.000008759419,0.8362065,0.001308754,0.1605113,0.001738522,0.00001802156],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.005575958,0.0002891063,0.9915803,0.0002226598,0.00004880735,0.00005789144,0.00007231512,0.0006352843,0.001517605],"genre_scores_gemma":[0.5346224,0.0007420531,0.4580865,0.0004122108,0.000150693,0.0008540702,0.0003211505,0.0002973563,0.004513677],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004578013,"threshold_uncertainty_score":0.015315,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.007398034949888881,"score_gpt":0.2720153161221875,"score_spread":0.2646172811722987,"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."}}