{"id":"W3123086775","doi":"10.1007/s10664-021-09982-4","title":"Can Offline Testing of Deep Neural Networks Replace Their Online Testing?","year":2021,"lang":"en","type":"preprint","venue":"Empirical Software Engineering","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Ottawa","funders":"H2020 European Research Council; Canada Research Chairs; Ministry of Education; National Research Foundation of Korea; Fonds National de la Recherche Luxembourg; National Research Foundation; Natural Sciences and Engineering Research Council of Canada; European Commission","keywords":"Computer science; Online and offline; Test strategy; Manual testing; Orthogonal array testing; White-box testing; Context (archaeology); Black-box testing; Exploit; Model-based testing; Machine learning; Artificial intelligence; Test case; Computer security; Operating system; Software","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":["metaresearch","metaepi_narrow","research_integrity"],"consensus_categories":[],"category_scores_codex":[0.0005815339,0.0007324307,0.00102424,0.0002244287,0.0001364167,0.0002291751,0.00202893,0.0005134956,0.0000112557],"category_scores_gemma":[0.01602037,0.0007376006,0.0002700669,0.001490249,0.00005888531,0.0002097766,0.004925348,0.003011558,6.539402e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000238084,"about_ca_system_score_gemma":0.0002080832,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001880452,"about_ca_topic_score_gemma":0.0000241711,"domain_scores_codex":[0.9961836,0.0001751383,0.001001788,0.001276819,0.0005523619,0.0008102828],"domain_scores_gemma":[0.993467,0.003659745,0.0005561672,0.001535908,0.0004973536,0.0002837669],"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":[0.000003297648,0.00006467423,0.05253236,0.0002136478,0.00005648121,0.0001043966,0.0003463369,0.9239166,0.00003508817,0.000009757418,0.00001807411,0.02269925],"study_design_scores_gemma":[0.0002150621,0.00006277485,0.02293465,0.0004361886,0.00003068081,0.0000675534,0.00001868743,0.9753945,0.00005668167,0.00006698829,0.00006459967,0.0006515568],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.110195,0.0007748669,0.8853611,0.0003610809,0.001648307,0.0002294559,0.00001280337,0.001401677,0.00001576861],"genre_scores_gemma":[0.5098892,0.000002579907,0.4892065,0.0001393552,0.0005971879,0.00001440909,0.00006449842,0.00007599666,0.00001021031],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.3996943,"threshold_uncertainty_score":0.9995075,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03600581497936919,"score_gpt":0.2730723730456751,"score_spread":0.2370665580663059,"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."}}