{"id":"W4283794091","doi":"10.1609/aaai.v36i1.19938","title":"PatchUp: A Feature-Space Block-Level Regularization Technique for Convolutional Neural Networks","year":2022,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":33,"is_retracted":false,"has_abstract":true,"ca_institutions":"Polytechnique Montréal; McGill University; Université de Montréal","funders":"Compute Canada; Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research","keywords":"Computer science; Convolutional neural network; Artificial intelligence; Regularization (linguistics); Robustness (evolution); Pattern recognition (psychology); Feature vector; Generalization; Block (permutation group theory); Deep neural networks; Deep learning; Machine learning; Mathematics","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":[],"consensus_categories":[],"category_scores_codex":[0.0009619723,0.000245213,0.0002598408,0.0001496254,0.0008473295,0.0001738647,0.002016864,0.0001047039,0.00003452579],"category_scores_gemma":[0.0005817566,0.0002174902,0.0001617236,0.0009306215,0.0001991193,0.000349827,0.001044476,0.0007153015,0.000002329019],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001358139,"about_ca_system_score_gemma":0.0001295384,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002223876,"about_ca_topic_score_gemma":0.000003160838,"domain_scores_codex":[0.9978739,0.00005849754,0.0004215464,0.0005974994,0.0006396911,0.0004088869],"domain_scores_gemma":[0.9982222,0.0001925413,0.0005600137,0.0003186689,0.0006352514,0.00007135512],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000113831,0.000085832,0.0001213762,0.00002017603,0.00001386459,5.154499e-7,0.0004457131,0.08598807,0.01597377,0.8908957,0.0004866471,0.005854474],"study_design_scores_gemma":[0.00004832556,0.0002133724,0.00006777381,0.00003947511,0.00001287526,0.00001767815,0.0002214642,0.8295805,0.0644018,0.1048872,0.0002765879,0.0002329523],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.002718758,0.00001974011,0.985621,0.008827554,0.0007788663,0.001067534,0.00001554273,0.0001273376,0.000823644],"genre_scores_gemma":[0.9652853,0.00000290143,0.03320398,0.0002758979,0.000137075,0.0003588484,0.000004260252,0.00002141427,0.0007103438],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9625665,"threshold_uncertainty_score":0.8868995,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05613656257909424,"score_gpt":0.2852592275984625,"score_spread":0.2291226650193683,"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."}}