{"id":"W4401687215","doi":"10.1109/access.2024.3446310","title":"Finding the Optimal Security Policies for Autonomous Cyber Operations With Competitive Reinforcement Learning","year":2024,"lang":"en","type":"article","venue":"IEEE Access","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"ca_institutions":"Defence Research and Development Canada; Royal Military College of Canada","funders":"","keywords":"Reinforcement learning; Computer science; Computer security; Artificial intelligence","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":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0003421033,0.000180078,0.000144072,0.0001199181,0.0006644335,0.002344514,0.001151688,0.00004622568,0.00002555691],"category_scores_gemma":[0.00004537628,0.0001209389,0.00006836541,0.0004204905,0.00009447811,0.0012673,0.0002767412,0.0003334601,0.00003402992],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001201213,"about_ca_system_score_gemma":0.0001827113,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001113929,"about_ca_topic_score_gemma":0.00003090517,"domain_scores_codex":[0.9986932,0.00005231023,0.0002444848,0.0003323001,0.0003028228,0.0003748938],"domain_scores_gemma":[0.9990653,0.0003359739,0.00005920442,0.0003552205,0.0001272756,0.00005698793],"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":[0.000004029276,0.000004693665,0.000100248,0.00002594104,0.00005112262,0.000004897999,0.003775695,0.8819544,0.00003927032,0.1133103,0.0004216278,0.0003078177],"study_design_scores_gemma":[0.0001774594,0.0001675792,0.0001316956,0.00009056577,0.00002211437,0.00001527668,0.0002118616,0.9776276,0.00161589,0.00008833136,0.01965617,0.0001954137],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01000819,0.00005137606,0.980958,0.001673329,0.0005767553,0.0005313461,0.000001817137,0.0003247346,0.005874417],"genre_scores_gemma":[0.9906484,0.00001403771,0.005923388,0.0003738912,0.0001958042,0.0001433113,0.00001043214,0.00002074344,0.002669934],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9806403,"threshold_uncertainty_score":0.9986911,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03052046548776626,"score_gpt":0.3178122432086103,"score_spread":0.287291777720844,"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."}}