{"id":"W4409381863","doi":"10.1609/aaai.v39i28.35158","title":"Exploring the Efficacy of Multi-Agent Reinforcement Learning for Autonomous Cyber Defence: A CAGE Challenge 4 Perspective","year":2025,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Advanced Malware Detection Techniques","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"ca_institutions":"Defence Research and Development Canada","funders":"","keywords":"Perspective (graphical); Reinforcement learning; Reinforcement; Cage; Computer science; Psychology; Cognitive science; Human–computer interaction; Engineering; Social psychology; Artificial intelligence; Structural engineering","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00389193,0.0004584829,0.0006056126,0.0003373382,0.0003587394,0.001043587,0.000872154,0.001251318,0.0016168],"category_scores_gemma":[0.01135539,0.0002256426,0.0003504181,0.0001340909,0.001431141,0.001976737,0.001264961,0.001670746,0.0001728858],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008691247,"about_ca_system_score_gemma":0.001007052,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00202388,"about_ca_topic_score_gemma":0.001565816,"domain_scores_codex":[0.9989665,0.0006394557,0.0000289178,0.0001278921,0.0001382171,0.00009899641],"domain_scores_gemma":[0.988874,0.009002507,0.0006308234,0.0005461082,0.0006114535,0.0003351785],"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":[0.0002080112,0.0002507377,0.002752869,0.0001247001,0.00006208655,0.00005449853,0.00009069303,0.9570978,0.002734225,0.01894668,0.0006794205,0.0169983],"study_design_scores_gemma":[0.00002089656,0.0002482008,0.0002501691,0.000008041314,0.000006003135,0.00001018091,0.00003489771,0.991725,0.0006522601,0.006719946,0.0003185943,0.000005811406],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5800608,0.001091568,0.3805922,0.0067829,0.0001585865,0.0003338675,0.0001065206,0.0004801844,0.03039346],"genre_scores_gemma":[0.9788114,0.0001177496,0.01980812,0.0001825764,0.00002558688,0.00006841159,0.00001983541,0.00001635036,0.0009499892],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.00389193,"threshold_uncertainty_score":0.02058274,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1502720827387915,"score_gpt":0.3429535311925003,"score_spread":0.1926814484537089,"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."}}