{"id":"W3203863013","doi":"","title":"Physical Security of Deep Learning on Edge Devices: Comprehensive Evaluation of Fault Injection Attack Vectors","year":2019,"lang":"en","type":"preprint","venue":"IACR Cryptology ePrint Archive","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Alberta","funders":"","keywords":"Softmax function; Computer science; Deep learning; Fault (geology); Enhanced Data Rates for GSM Evolution; Computer security; Artificial intelligence; Countermeasure; Artificial neural network; Fault injection; Attack surface; Deep neural networks; Edge device; Reliability (semiconductor); Machine learning; Software; Power (physics); Cloud computing","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":[0.001311496,0.0007048166,0.0004047511,0.0005713876,0.0001720839,0.0003979455,0.0005111059,0.000667869,0.001041549],"category_scores_gemma":[0.004756813,0.0001310249,0.0003108096,0.0003037109,0.0007131939,0.0008596654,0.0006371648,0.0006704538,0.0001256371],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007199942,"about_ca_system_score_gemma":0.0002522836,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001100259,"about_ca_topic_score_gemma":0.0007504044,"domain_scores_codex":[0.9992906,0.0002413612,0.00005134231,0.00008164313,0.0002219091,0.0001131172],"domain_scores_gemma":[0.996187,0.002503914,0.0004073005,0.0004099249,0.0003662843,0.0001255818],"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":[0.0008253159,0.0002615875,0.004243277,0.0001716415,0.0001014311,0.0002016543,0.00003502646,0.9447314,0.01026725,0.002076758,0.0009628045,0.03612185],"study_design_scores_gemma":[0.00001514862,0.000599126,0.001096247,0.00001951545,0.00002560325,0.00006489907,0.00002116072,0.9794853,0.01754061,0.0009005026,0.0002239177,0.000007933001],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9430218,0.000710008,0.05250383,0.000344104,0.00008798617,0.00005728679,0.0001586452,0.000602386,0.002513826],"genre_scores_gemma":[0.9963965,0.0000756895,0.003124913,0.00003591349,0.000003233479,0.000008017756,0.00005614118,0.00001207399,0.0002875704],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.001311496,"threshold_uncertainty_score":0.006935954,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03799752205158395,"score_gpt":0.3391170141051966,"score_spread":0.3011194920536127,"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."}}