{"id":"W4417077291","doi":"10.48550/arxiv.2505.23027","title":"Diverse Prototypical Ensembles Improve Robustness to Subpopulation Shift","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Domain Adaptation and Few-Shot Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Canadian Institutes of Health Research; Natural Sciences and Engineering Research Council of Canada; Government of Canada; Canadian Institute for Advanced Research","keywords":"Robustness (evolution); Training set; Minification; Ensemble learning; Feature (linguistics); Pattern recognition (psychology); Extractor; Empirical risk minimization; Support vector machine","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003957523,0.001216762,0.001599738,0.001076094,0.001011092,0.001238808,0.001812074,0.001600909,0.001029539],"category_scores_gemma":[0.01563601,0.0004906887,0.000933457,0.00063685,0.0009106706,0.002893717,0.003438375,0.002406559,0.0006626488],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005791782,"about_ca_system_score_gemma":0.0008257854,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001781626,"about_ca_topic_score_gemma":0.003117355,"domain_scores_codex":[0.9981561,0.0005832571,0.000111343,0.0006166286,0.0003612815,0.000171302],"domain_scores_gemma":[0.9943134,0.00237899,0.0003484376,0.001818759,0.0008278499,0.0003125931],"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":[0.0003812779,0.0002966577,0.0173849,0.0001282612,0.0003896003,0.0001984619,0.0005936726,0.5729216,0.0158322,0.00913626,0.009316167,0.3734209],"study_design_scores_gemma":[0.00001337593,0.00008029219,0.001116944,0.0000149306,0.00002703177,0.00006781053,0.00006259104,0.9862593,0.003368692,0.007821091,0.001152438,0.00001542523],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.2270885,0.001156734,0.7657412,0.0005099233,0.0001450376,0.0001048554,0.0001771976,0.002273264,0.002803312],"genre_scores_gemma":[0.8833106,0.0002381797,0.1119671,0.0005524087,0.0001382992,0.0001318819,0.0008208368,0.0002954497,0.002545326],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003957523,"threshold_uncertainty_score":0.02092963,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05673509766038131,"score_gpt":0.3015717884706264,"score_spread":0.2448366908102451,"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."}}