{"id":"W2964855005","doi":"10.1609/aiide.v15i1.5220","title":"On Hard Exploration for Reinforcement Learning: A Case Study in Pommerman","year":2019,"lang":"en","type":"preprint","venue":"Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Alberta","funders":"","keywords":"Reinforcement learning; SAFER; Benchmark (surveying); Pruning; Computer science; Domain (mathematical analysis); Artificial intelligence; Machine learning; Computer security; Mathematics","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":["metaepi_narrow","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0004155624,0.0004641125,0.0004886325,0.0003360205,0.0001360581,0.001112895,0.00109294,0.0001259013,0.000009278445],"category_scores_gemma":[0.0004510971,0.0003690294,0.000186737,0.0001829221,0.00009406785,0.001060794,0.001436966,0.0008545113,0.0000216929],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002865164,"about_ca_system_score_gemma":0.00008491845,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006507942,"about_ca_topic_score_gemma":0.000007158188,"domain_scores_codex":[0.9972016,0.00003307013,0.0009362661,0.0008715536,0.0005857497,0.0003717155],"domain_scores_gemma":[0.997892,0.0002598853,0.0009066947,0.0003900403,0.0004711372,0.00008023074],"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":[0.001225471,0.002159118,0.003319388,0.000497966,0.0003525896,0.00003821356,0.04925877,0.6932538,0.0006751228,0.1754266,0.0001499362,0.07364307],"study_design_scores_gemma":[0.0002501126,0.004340426,0.00007851253,0.001360126,0.00003039875,0.0000182685,0.02051544,0.9469361,0.01242779,0.01342634,0.00008825996,0.0005282559],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6216235,0.000006368068,0.3643009,0.001375643,0.001045858,0.005140146,0.000009083136,0.00006843585,0.006430154],"genre_scores_gemma":[0.9984994,0.00001985407,0.0001369749,0.0001529425,0.00003361785,0.0003284265,0.000008486869,0.000023273,0.0007970565],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3768759,"threshold_uncertainty_score":0.9999241,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1009064224702263,"score_gpt":0.3290656216999939,"score_spread":0.2281591992297676,"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."}}