{"id":"W3174355057","doi":"10.1609/aaai.v35i11.17163","title":"DIBS: Diversity Inducing Information Bottleneck in Model Ensembles","year":2021,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal; Vector Institute; University of Toronto","funders":"","keywords":"MNIST database; Computer science; Generalization; Machine learning; Artificial intelligence; Benchmark (surveying); Information bottleneck method; Bayesian probability; Bottleneck; Artificial neural network; Data mining; Mutual information; Mathematics; Geography","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":[0.005081934,0.002039139,0.002514223,0.0009746546,0.0009560829,0.00161417,0.003206262,0.001965633,0.002405756],"category_scores_gemma":[0.01451284,0.001074213,0.001475261,0.0009577902,0.001416423,0.0035693,0.00510316,0.004780564,0.0009273253],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001637329,"about_ca_system_score_gemma":0.002054441,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004105334,"about_ca_topic_score_gemma":0.004610647,"domain_scores_codex":[0.9977213,0.0008835623,0.0001183709,0.0004424275,0.0006122452,0.0002220931],"domain_scores_gemma":[0.9939354,0.003378258,0.0004487614,0.001213096,0.0006397832,0.0003847288],"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":[0.000255062,0.0001478076,0.001613124,0.0001244685,0.000170561,0.0001585755,0.0001344856,0.8616834,0.003496365,0.01775187,0.005527293,0.108937],"study_design_scores_gemma":[0.000009197842,0.00002865582,0.00006466373,0.000009477721,0.000008708682,0.00002205299,0.000006111578,0.9901457,0.001049295,0.008291015,0.0003592012,0.000005894837],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02910574,0.0007526344,0.9653918,0.0005275796,0.00009353756,0.00008114448,0.0001814292,0.002274457,0.001591735],"genre_scores_gemma":[0.7382812,0.0006372322,0.2527112,0.001098026,0.0002692405,0.000438525,0.001469825,0.0008293532,0.004265443],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.005081934,"threshold_uncertainty_score":0.02687615,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07016165741512634,"score_gpt":0.2809434166131086,"score_spread":0.2107817591979823,"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."}}