{"id":"W4415589112","doi":"10.1145/3773034","title":"Synthesizing Efficient and Permissive Programmatic Runtime Shields for Neural Policies","year":2025,"lang":"en","type":"article","venue":"ACM Transactions on Software Engineering and Methodology","topic":"Neural Networks and Applications","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Alberta Medical Association","funders":"","keywords":"Shields; Overhead (engineering); Reduction (mathematics); Software; Control (management); Reliability (semiconductor); Safety standards; Runtime system","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":[],"consensus_categories":[],"category_scores_codex":[0.0002301202,0.0001200099,0.0001773145,0.0001345757,0.0002000191,0.00005780819,0.0002130244,0.00007131074,0.000001061354],"category_scores_gemma":[0.0002753414,0.0001095713,0.0000478175,0.0002208124,0.00003850189,0.0000462984,0.00001960278,0.0001368545,3.345699e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001034257,"about_ca_system_score_gemma":0.00001344678,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000008994184,"about_ca_topic_score_gemma":0.000001285132,"domain_scores_codex":[0.9993126,0.00003766741,0.0001378018,0.0002609234,0.00004066237,0.0002103086],"domain_scores_gemma":[0.9970838,0.002528516,0.00002392828,0.0002790312,0.00002723347,0.00005748437],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00001592652,0.00006262936,0.00004815214,0.0001985339,0.00007853546,0.000001757459,0.0007565314,0.1495576,0.001906628,0.02835367,0.0000422547,0.8189778],"study_design_scores_gemma":[0.0007557392,0.0003176903,0.003033543,0.0001858132,0.0001517624,0.0001036169,0.0001597792,0.9727607,0.00992786,0.005561787,0.006503087,0.0005385556],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.03431465,0.0003118405,0.963209,0.00147032,0.0001952337,0.0002577362,0.000003182482,0.0002351652,0.000002901244],"genre_scores_gemma":[0.2730232,0.00003467748,0.7264389,0.0001928947,0.00001930322,0.0001897929,6.198102e-7,0.000007824611,0.00009281142],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.8232031,"threshold_uncertainty_score":0.4468188,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0516818758031873,"score_gpt":0.3185185259233981,"score_spread":0.2668366501202108,"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."}}