{"id":"W2895478303","doi":"10.48550/arxiv.1810.01032","title":"Reinforcement Learning with Perturbed Rewards","year":2018,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Reinforcement learning; Computer science; Noise (video); Confusion matrix; Artificial intelligence; Convergence (economics); Confusion; Set (abstract data type); Gaussian; Machine learning; Matrix (chemical analysis); Mathematical optimization; Algorithm; 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"],"consensus_categories":[],"category_scores_codex":[0.0003975696,0.0004979665,0.0004212342,0.0003304305,0.0003545504,0.000311066,0.002578481,0.0003317852,0.0001126057],"category_scores_gemma":[0.0000833095,0.0004991131,0.0001865986,0.0006326672,0.0002233149,0.0004920528,0.003182948,0.001191131,0.0003194128],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004167123,"about_ca_system_score_gemma":0.0003838235,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00008670337,"about_ca_topic_score_gemma":0.000007067467,"domain_scores_codex":[0.9972911,0.0001730716,0.0002965937,0.001320509,0.0002948259,0.0006239256],"domain_scores_gemma":[0.9969497,0.0001017456,0.00052582,0.00180662,0.0003772959,0.000238817],"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":[0.0000397314,0.00001550544,0.001876704,0.00006421701,0.0001351838,0.000185728,0.0004283346,0.9748817,0.000006180671,0.02180403,0.0004518949,0.0001107849],"study_design_scores_gemma":[0.0005827078,0.0004385648,0.0003356065,0.0002109789,0.00007465584,0.000009283199,0.00006782396,0.9914374,0.0000897249,0.0008166537,0.005272031,0.0006645815],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01850641,0.00001709449,0.9587017,0.00009589128,0.0005409274,0.0003598718,5.822206e-7,0.0006156074,0.02116191],"genre_scores_gemma":[0.9703809,0.00009656811,0.006435538,0.0001476389,0.0001442446,0.000001603997,0.00002639281,0.00003827525,0.02272887],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9522662,"threshold_uncertainty_score":0.999746,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05333297638058677,"score_gpt":0.185023905452045,"score_spread":0.1316909290714582,"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."}}