{"id":"W4411449759","doi":"10.1145/3715736","title":"One-for-All Does Not Work! Enhancing Vulnerability Detection by Mixture-of-Experts (MoE)","year":2025,"lang":"en","type":"article","venue":"Proceedings of the ACM on software engineering.","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"ca_institutions":"Huawei Technologies (Canada); University of Manitoba","funders":"","keywords":"Vulnerability (computing); Computer science; Task (project management); Artificial intelligence; Baseline (sea); Deep learning; Machine learning; Code (set theory); Computer security; Engineering; Biology","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005310628,0.003987368,0.002365205,0.00276934,0.0009540166,0.002512401,0.00328479,0.003991501,0.004511975],"category_scores_gemma":[0.01365912,0.001232172,0.003011136,0.001245838,0.001488747,0.009675007,0.005180401,0.00538598,0.005707867],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001248832,"about_ca_system_score_gemma":0.001846194,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004856099,"about_ca_topic_score_gemma":0.01003023,"domain_scores_codex":[0.995644,0.001140169,0.0002438214,0.001431249,0.001018901,0.0005218375],"domain_scores_gemma":[0.9948208,0.002048858,0.0003503031,0.001662679,0.000766433,0.0003508123],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0009883525,0.0006475639,0.02145188,0.0007731168,0.001016319,0.0005903561,0.0003671245,0.0931206,0.01545807,0.008363274,0.07641772,0.7808056],"study_design_scores_gemma":[0.00008788864,0.0004056105,0.002707162,0.0002301424,0.0002404575,0.001273477,0.0002185164,0.9068868,0.02095887,0.03650761,0.03032805,0.0001555353],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.08654629,0.009282838,0.8327342,0.005954359,0.001049066,0.0003132099,0.002279264,0.05105182,0.010789],"genre_scores_gemma":[0.526357,0.002495515,0.4408609,0.005691069,0.0003505616,0.0002578227,0.006337742,0.002831576,0.01481772],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005310628,"threshold_uncertainty_score":0.02808559,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.008965197819905181,"score_gpt":0.2428457572716993,"score_spread":0.2338805594517941,"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."}}