{"id":"W4400351643","doi":"10.1109/tse.2024.3423712","title":"Revisiting the Performance of Deep Learning-Based Vulnerability Detection on Realistic Datasets","year":2024,"lang":"en","type":"article","venue":"IEEE Transactions on Software Engineering","topic":"Network Security and Intrusion Detection","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Vulnerability (computing); Artificial intelligence; Deep learning; Data science; Machine learning; Computer security","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":[],"consensus_categories":[],"category_scores_codex":[0.0004666841,0.0001478057,0.0001229649,0.0001724497,0.000277421,0.00009636647,0.0002626897,0.00006878704,0.00002142315],"category_scores_gemma":[0.00005865927,0.0001226592,0.0000933086,0.000661991,0.00002493591,0.0002814608,0.000002982423,0.0006618651,0.00002054346],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009162749,"about_ca_system_score_gemma":0.00001865007,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001683226,"about_ca_topic_score_gemma":0.00000539283,"domain_scores_codex":[0.998935,0.00007364991,0.0002413237,0.0003098738,0.0002548508,0.0001852843],"domain_scores_gemma":[0.9988531,0.0006071308,0.00004105834,0.000415765,0.00003817474,0.00004478873],"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":[0.0000130595,0.00001044619,0.000002107472,0.0001169322,0.000009950309,0.000001776324,0.00006523359,0.7569088,0.0003086404,0.00004085413,0.000002973511,0.2425192],"study_design_scores_gemma":[0.00006903216,0.0001747902,0.000180451,0.0002280433,0.00001388815,0.00001032709,0.000003010322,0.9492576,0.04859672,0.000006821553,0.001339979,0.0001194003],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04398535,0.00007893289,0.9540475,0.00005345881,0.001078556,0.0001149139,0.00001203678,0.0006233879,0.000005875068],"genre_scores_gemma":[0.9973282,0.00003043134,0.002439669,0.0000296692,0.0001115662,0.0000320952,0.000003832015,0.00001615354,0.00000833668],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9533429,"threshold_uncertainty_score":0.5001896,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.009199555430048047,"score_gpt":0.2236496439675864,"score_spread":0.2144500885375384,"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."}}