{"id":"W3010028534","doi":"10.1109/cvpr42600.2020.00132","title":"Learn2Perturb: An End-to-End Feature Perturbation Learning to Improve Adversarial Robustness","year":2020,"lang":"en","type":"preprint","venue":"","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":4,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Adversarial system; Deep neural networks; Leverage (statistics); Computer science; Artificial neural network; Artificial intelligence; Robustness (evolution); Deep learning; Machine learning; Perturbation (astronomy); End-to-end principle; Inference","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":[0.001708142,0.00183692,0.001350317,0.0006022049,0.0004506831,0.0007819165,0.002119081,0.001729767,0.003021031],"category_scores_gemma":[0.005930993,0.0005310357,0.0008072646,0.0003575631,0.001577571,0.002354487,0.003442554,0.003341977,0.001150097],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000735475,"about_ca_system_score_gemma":0.001008475,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0009889939,"about_ca_topic_score_gemma":0.001367864,"domain_scores_codex":[0.998843,0.0002693133,0.00005766069,0.0002974697,0.0004039157,0.0001287046],"domain_scores_gemma":[0.9984468,0.0006190413,0.0001655892,0.0004275682,0.0002401567,0.0001009316],"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":[0.0003556623,0.0002483163,0.001261384,0.0001213879,0.0001277107,0.0002575861,0.00009815412,0.7365919,0.02052657,0.01170866,0.008714533,0.2199883],"study_design_scores_gemma":[0.00001236079,0.00008916148,0.00009810438,0.000006079331,0.000008162585,0.00004805632,0.00000571231,0.9894818,0.005257458,0.004237233,0.0007466617,0.000009092518],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01252318,0.0002442081,0.9809471,0.0001925727,0.00008443067,0.0001154866,0.00008955658,0.004541304,0.001262149],"genre_scores_gemma":[0.6159272,0.0003474902,0.3739653,0.0008778764,0.0001569343,0.000375512,0.0008187141,0.0008990624,0.006631993],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003021031,"threshold_uncertainty_score":0.01010638,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01896137194560248,"score_gpt":0.2761250068037797,"score_spread":0.2571636348581772,"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."}}