{"id":"W3116666954","doi":"10.1109/spw50608.2020.00027","title":"On the Robustness of Cooperative Multi-Agent Reinforcement Learning","year":2020,"lang":"en","type":"article","venue":"","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":56,"is_retracted":false,"has_abstract":true,"ca_institutions":"Vector Institute; University of Toronto","funders":"Canadian Institute for Advanced Research","keywords":"Reinforcement learning; Adversary; Robustness (evolution); Adversarial system; Computer science; Harm; Artificial intelligence; Computer security; Benchmark (surveying); Differentiable function; Mathematics; Psychology; Social psychology","routes":{"ca_aff":true,"ca_fund":true,"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":[],"consensus_categories":[],"category_scores_codex":[0.0003097683,0.0001367538,0.0001629927,0.00003056215,0.0002041359,0.00005948617,0.0008515262,0.00003479039,0.000237957],"category_scores_gemma":[0.000699567,0.00008586175,0.00005845296,0.0003676316,0.00005537645,0.0001832493,0.0004644582,0.0003737865,0.00005646819],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000276304,"about_ca_system_score_gemma":0.00004729377,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001767361,"about_ca_topic_score_gemma":0.00000108313,"domain_scores_codex":[0.998751,0.0002046715,0.0002412133,0.0002841365,0.0003309269,0.000188012],"domain_scores_gemma":[0.9990562,0.0003442161,0.000135421,0.0002915258,0.00009939433,0.00007324624],"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":[0.000009033784,0.00001329704,0.00005798618,0.000005747081,0.00001595223,0.000002382534,0.00142698,0.8977878,0.0001566104,0.09906721,0.0003635401,0.001093464],"study_design_scores_gemma":[0.0003209389,0.0001864853,0.00006584654,0.00001412311,0.000003542719,6.761561e-7,0.0002892986,0.996285,0.001875013,0.00001710629,0.00083333,0.0001086913],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.00371152,0.000008135622,0.9863191,0.00573994,0.0001515177,0.0002066445,1.115042e-7,0.0001257549,0.003737253],"genre_scores_gemma":[0.9664438,0.000002910733,0.03126853,0.001499878,0.00005018068,0.00001025526,0.000001028293,0.000009090903,0.0007143529],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9627323,"threshold_uncertainty_score":0.350134,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04048919975561101,"score_gpt":0.27154417538606,"score_spread":0.231054975630449,"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."}}