{"id":"W3173019690","doi":"10.1109/tpami.2022.3174724","title":"Adversarial Robustness Via Fisher-Rao Regularization","year":2022,"lang":"en","type":"article","venue":"IEEE Transactions on Pattern Analysis and Machine Intelligence","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":26,"is_retracted":false,"has_abstract":true,"ca_institutions":"École de Technologie Supérieure; McGill University","funders":"Grand Équipement National De Calcul Intensif; European Commission; Natural Sciences and Engineering Research Council of Canada; McGill University","keywords":"Softmax function; Robustness (evolution); Computer science; Artificial intelligence; Regularization (linguistics); Geodesic; Categorical variable; Entropy (arrow of time); Gaussian; Binary number; Pattern recognition (psychology); Algorithm; Mathematics; Machine learning; Artificial neural network","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003074896,0.001482358,0.001128994,0.001034934,0.0004516732,0.001254394,0.001324961,0.001409861,0.001920884],"category_scores_gemma":[0.009731623,0.0003843945,0.0008994593,0.0006587276,0.002145614,0.001791029,0.002623455,0.002492688,0.000756164],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001112182,"about_ca_system_score_gemma":0.000851721,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001019277,"about_ca_topic_score_gemma":0.000776063,"domain_scores_codex":[0.9981321,0.0006760454,0.00007579828,0.0002876683,0.0006667263,0.0001617297],"domain_scores_gemma":[0.9962846,0.002401311,0.000390324,0.0005048765,0.0003073525,0.0001115094],"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":[0.00009660944,0.0000347671,0.0005681189,0.00006625113,0.0000608678,0.00008780666,0.00004530862,0.8960019,0.005731469,0.06350365,0.002225291,0.0315779],"study_design_scores_gemma":[0.000002929591,0.00002889035,0.0001236558,0.000009775114,0.000004432169,0.00004091523,0.000005016479,0.9780212,0.001610847,0.0196081,0.0005337993,0.00001045276],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01095259,0.0002842791,0.9851842,0.0002805441,0.00003639044,0.00002531798,0.00006498567,0.0003441264,0.00282767],"genre_scores_gemma":[0.8186525,0.0007774116,0.1711152,0.0004544158,0.000158719,0.0002168267,0.0004710361,0.0003282573,0.00782565],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003074896,"threshold_uncertainty_score":0.01626182,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01440254615196882,"score_gpt":0.2477010985730526,"score_spread":0.2332985524210837,"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."}}