{"id":"W3217650841","doi":"10.1007/978-3-031-17143-7_19","title":"Real-Time Adversarial Perturbations Against Deep Reinforcement Learning Policies: Attacks and Defenses","year":2022,"lang":"en","type":"book-chapter","venue":"Lecture notes in computer science","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Adversarial system; Computer science; Reinforcement learning; Artificial intelligence; Computer security","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","sts"],"consensus_categories":[],"category_scores_codex":[0.001107754,0.0006962152,0.0006709991,0.001093725,0.001403599,0.0006645942,0.002748576,0.000299464,0.0001846675],"category_scores_gemma":[0.000513324,0.0007139414,0.0001636545,0.0007530536,0.0007741685,0.0009810366,0.004215036,0.001829143,0.00003542038],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007599688,"about_ca_system_score_gemma":0.0006290473,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001551045,"about_ca_topic_score_gemma":0.00002988631,"domain_scores_codex":[0.9950937,0.0001452049,0.0006973001,0.001745399,0.001428699,0.0008896665],"domain_scores_gemma":[0.9966003,0.001239386,0.0005405222,0.001176803,0.0001972788,0.0002457374],"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":[0.000009562177,0.000009906205,0.00006361353,0.0000204387,0.00001951894,0.00005556899,0.002920052,0.8842263,0.0001276862,0.01815988,0.00001445699,0.09437305],"study_design_scores_gemma":[0.0004444067,0.0002494264,0.00008462732,0.0001411996,0.0000190395,0.00005523538,0.000002480699,0.984848,0.00003578604,0.007905551,0.005397792,0.000816447],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.0002900858,0.0001627873,0.9773791,0.0007408525,0.001362974,0.0004603523,0.000001945597,0.000363685,0.01923826],"genre_scores_gemma":[0.3360343,0.001017455,0.6502518,0.003315541,0.002410051,0.00007576006,0.0001184238,0.0002514432,0.006525182],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.3357442,"threshold_uncertainty_score":0.9998965,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01209706532778705,"score_gpt":0.249908953409619,"score_spread":0.2378118880818319,"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."}}