{"id":"W2981902149","doi":"10.48550/arxiv.1910.10307","title":"Detecting Out-of-Distribution Inputs in Deep Neural Networks Using an Early-Layer Output","year":2019,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Anomaly Detection Techniques and Applications","field":"Computer Science","cited_by":35,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Layer (electronics); Artificial neural network; Distribution (mathematics); Computer science; Deep neural networks; Artificial intelligence; Mathematics; Materials science; Nanotechnology","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":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0002347395,0.0002320919,0.0002867938,0.0002121638,0.0001193286,0.00008501476,0.001067081,0.0003476856,0.00000356241],"category_scores_gemma":[0.00001166829,0.0002910569,0.000134113,0.0006017469,0.00005641925,0.0004406198,0.001068043,0.0006398763,0.000006058245],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002465061,"about_ca_system_score_gemma":0.00005636734,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003749946,"about_ca_topic_score_gemma":0.0001181277,"domain_scores_codex":[0.9983878,0.000126807,0.0002824853,0.0008230454,0.00007373818,0.0003060903],"domain_scores_gemma":[0.9983424,0.0000472028,0.0003817284,0.0009998044,0.0001333918,0.00009544961],"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":[0.00001086845,0.00007050214,0.01070036,0.00002123958,0.00001363464,0.00001752124,0.000125123,0.9816906,0.0001359164,0.004529138,0.000003639694,0.002681396],"study_design_scores_gemma":[0.0001574678,0.00005897253,0.004247492,0.00003467884,0.00001955182,0.000001969603,0.00002157398,0.9924873,0.000592742,0.002073355,0.00002625594,0.0002786683],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.450197,0.00000957168,0.5491955,0.000008174678,0.0002140784,0.0001992253,0.000005475809,0.0001159894,0.00005492976],"genre_scores_gemma":[0.9973463,0.00001336149,0.002458695,0.00002963023,0.00005741701,0.000001649767,0.00002194801,0.00001492957,0.00005606859],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.5471492,"threshold_uncertainty_score":0.9999542,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.09811232229998548,"score_gpt":0.2244071822298095,"score_spread":0.126294859929824,"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."}}