{"id":"W3214544808","doi":"10.48550/arxiv.2002.08313","title":"NNoculation: Catching BadNets in the Wild","year":2020,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Calgary","funders":"","keywords":"Backdoor; Software deployment; Computer science; Adversarial system; Field (mathematics); Computer security; Artificial intelligence; Contrast (vision); Suite; Machine learning; Mathematics; Software engineering","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":[0.0004905242,0.000272891,0.0002676715,0.0001745286,0.0001970173,0.0001931975,0.003354989,0.0002102496,0.00001540096],"category_scores_gemma":[0.0001220111,0.0002663247,0.0001473371,0.0009146326,0.00006425159,0.0004355737,0.002527415,0.001517835,0.00006202559],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001675065,"about_ca_system_score_gemma":0.0001674445,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003365313,"about_ca_topic_score_gemma":0.00006030164,"domain_scores_codex":[0.9978808,0.0004683556,0.0002043549,0.0009912987,0.0001579731,0.0002972285],"domain_scores_gemma":[0.9982622,0.0002587519,0.0002299084,0.00111675,0.00004986816,0.00008250793],"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":[0.000007125589,0.0000184798,0.005591142,0.00002564475,0.00001697998,0.0004493269,0.001527526,0.7820991,0.000002228952,0.2095032,0.0001225094,0.000636821],"study_design_scores_gemma":[0.0002802551,0.0000154487,0.007143408,0.00006322714,0.00002155134,0.000005619102,0.0001478057,0.9227585,0.000002096514,0.06842312,0.0008397096,0.000299254],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.05187218,0.00003988995,0.9398211,0.002667,0.0005751214,0.0002807189,0.000001782218,0.0001979547,0.004544218],"genre_scores_gemma":[0.9945384,0.00002152044,0.004551624,0.0005591132,0.0001724808,9.0722e-7,0.00001147946,0.00001475695,0.0001296948],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9426662,"threshold_uncertainty_score":0.9999789,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.09004250246331236,"score_gpt":0.2089939376827504,"score_spread":0.118951435219438,"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."}}