{"id":"W4390841592","doi":"","title":"A Geometric Perspective on ML Safety Assurance","year":2023,"lang":"en","type":"preprint","venue":"HAL (Le Centre pour la Communication Scientifique Directe)","topic":"Safety Systems Engineering in Autonomy","field":"Engineering","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Safran Electronics (Canada)","funders":"","keywords":"Perspective (graphical); Safety assurance; Safety case; Business; Computer science; Risk analysis (engineering); 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":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.00349387,0.0005275549,0.0005679694,0.0007480732,0.000172359,0.000238075,0.00128828,0.0004162979,0.00007333785],"category_scores_gemma":[0.002261112,0.0006268261,0.0002829054,0.00120582,0.00009336807,0.00009394042,0.0006228057,0.001167334,0.0005092291],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001094974,"about_ca_system_score_gemma":0.000170458,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000727355,"about_ca_topic_score_gemma":0.0003633737,"domain_scores_codex":[0.9962622,0.001233588,0.0006395357,0.0008366609,0.0005165555,0.0005114696],"domain_scores_gemma":[0.9933097,0.00260487,0.0002146263,0.00250501,0.001178539,0.0001871909],"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":[0.00004077933,0.0006262036,0.001897116,0.002682014,0.001493384,0.00006955607,0.02366661,0.691443,0.001853237,0.2227063,0.02054203,0.03297979],"study_design_scores_gemma":[0.002471556,0.000002003926,0.08705176,0.01707808,0.0002067302,0.00004689025,0.0008581647,0.7090653,0.03802317,0.00695141,0.1338317,0.004413174],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01505624,0.003492281,0.7329192,0.006852729,0.003282368,0.001445311,0.0003847493,0.006455034,0.2301121],"genre_scores_gemma":[0.9705714,0.001047777,0.01683566,0.00002994396,0.00009207752,0.0002508179,0.0002598284,0.0002718671,0.0106406],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9555152,"threshold_uncertainty_score":0.9996183,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01392351218285526,"score_gpt":0.2199947110849804,"score_spread":0.2060711989021251,"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."}}