{"id":"W4282827306","doi":"10.1109/icse-companion55297.2022.9793750","title":"HUDD: A tool to debug DNNs for safety analysis","year":2022,"lang":"en","type":"article","venue":"2022 IEEE/ACM 44th International Conference on Software Engineering: Companion Proceedings (ICSE-Companion)","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Ottawa","funders":"European Research Council; Natural Sciences and Engineering Research Council of Canada; Fonds National de la Recherche Luxembourg; European Commission","keywords":"Debugging; Computer science; Cluster analysis; Retraining; Root cause; Deep neural networks; Root (linguistics); Artificial neural network; Artificial intelligence; Image (mathematics); Domain (mathematical analysis); Pattern recognition (psychology); Machine learning; Reliability engineering; 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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002271215,0.002288311,0.0006030736,0.001866158,0.0003964871,0.001365191,0.00353017,0.001389589,0.02279799],"category_scores_gemma":[0.0109914,0.001165555,0.00107902,0.0004890171,0.0008719927,0.002336514,0.002736424,0.002569835,0.004161365],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009942558,"about_ca_system_score_gemma":0.001279853,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003204897,"about_ca_topic_score_gemma":0.006291052,"domain_scores_codex":[0.999235,0.0001830775,0.00009278519,0.0001927288,0.0002310583,0.00006532669],"domain_scores_gemma":[0.9964294,0.002519694,0.0002209598,0.0004555974,0.0002828381,0.00009158147],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00115303,0.0003167143,0.01177273,0.002452745,0.0005544146,0.002014959,0.00100959,0.238309,0.02155598,0.02330172,0.1856172,0.511942],"study_design_scores_gemma":[0.0002454373,0.0001467662,0.001272892,0.0002474076,0.00005807671,0.0004744025,0.0001280713,0.8648391,0.03585208,0.03245644,0.06417672,0.0001026031],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"software","genre_scores_codex":[0.007838158,0.0004693577,0.6931722,0.0003832314,0.0002493109,0.0001944046,0.004222516,0.2903883,0.003082594],"genre_scores_gemma":[0.1936218,0.0007826461,0.7484722,0.001003727,0.00007233272,0.001051197,0.01133906,0.03406822,0.009588868],"genre_candidate":"software","genre_consensus":null,"teacher_disagreement_score":0.02279799,"threshold_uncertainty_score":0.07626688,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02880629283998286,"score_gpt":0.278955330687586,"score_spread":0.2501490378476032,"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."}}