{"id":"W4392114235","doi":"10.1109/tvt.2024.3369100","title":"REAL-SAP: Real-Time Evidence Aware Liable Safety Assessment for Perception in Autonomous Driving","year":2024,"lang":"en","type":"article","venue":"IEEE Transactions on Vehicular Technology","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Perception; Reliability engineering; Computer science; Real-time computing; Engineering; 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":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0007496689,0.0002749648,0.0003409291,0.0009674087,0.0003406653,0.0001467344,0.0008546359,0.0004300976,0.00006145303],"category_scores_gemma":[0.00003388923,0.0002887673,0.0001635708,0.001377638,0.0001048041,0.0007374198,0.00001997806,0.0009128675,0.00009190144],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009396877,"about_ca_system_score_gemma":0.0002945556,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001900118,"about_ca_topic_score_gemma":0.0000866968,"domain_scores_codex":[0.997705,0.000132963,0.0004556413,0.0009182874,0.0002767626,0.0005113792],"domain_scores_gemma":[0.9985232,0.0004399958,0.00008021095,0.0008085339,0.00008621278,0.00006178847],"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":[0.00002889581,0.0001791193,0.0001810939,0.0001555928,0.00008438418,0.0001482842,0.0005012634,0.7316777,0.05582166,0.009506483,0.00008700119,0.2016285],"study_design_scores_gemma":[0.0002943936,0.0002663633,0.0005003656,0.0005036267,0.00003582274,0.00005217517,0.0001069918,0.9933502,0.002311467,0.001611914,0.0006488261,0.0003178531],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02649388,0.00004096015,0.9657526,0.004218853,0.0007433542,0.000603742,0.000005400128,0.001939925,0.000201278],"genre_scores_gemma":[0.8667275,0.0003134964,0.1320481,0.00003066966,0.00004424335,0.0002852426,0.000003446222,0.00004164679,0.0005055933],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8402337,"threshold_uncertainty_score":0.9999564,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01283765923528781,"score_gpt":0.2998226571443074,"score_spread":0.2869849979090197,"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."}}