{"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":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.0006319942,0.0002242334,0.0003274065,0.0001541003,0.0001173136,0.0000610629,0.002385683,0.0001443399,0.000001806557],"category_scores_gemma":[0.01123955,0.0001734431,0.0001636461,0.0006260306,0.00004536677,0.0003044647,0.0009454452,0.0003491605,5.143564e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001278178,"about_ca_system_score_gemma":0.00002623578,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001642501,"about_ca_topic_score_gemma":0.000001331475,"domain_scores_codex":[0.9984818,0.00001040062,0.0004060857,0.0004552167,0.0003465676,0.0002999899],"domain_scores_gemma":[0.9981262,0.0006425279,0.0002537738,0.0006631604,0.0002664887,0.00004784955],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0004572228,0.00045375,0.02149784,0.003030056,0.0004698931,4.187419e-7,0.003641606,0.04927856,0.7814587,0.01616769,0.002572766,0.1209715],"study_design_scores_gemma":[0.0004016797,0.00008615879,0.002766999,0.0003893642,0.00003419392,7.362049e-7,0.0000241849,0.01558004,0.9770063,0.001979068,0.001447841,0.0002834846],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2398011,0.00007050284,0.7575458,0.0007924061,0.001043674,0.0003639059,0.000003960476,0.0003462073,0.00003249044],"genre_scores_gemma":[0.8544341,0.00000288318,0.1451733,0.0001292329,0.00006474061,0.00005243121,5.765787e-7,0.00002040538,0.0001223298],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.614633,"threshold_uncertainty_score":0.9970892,"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."}}