{"id":"W4385571199","doi":"10.18653/v1/2023.repl4nlp-1.4","title":"A Multilingual Evaluation of NER Robustness to Adversarial Inputs","year":2023,"lang":"en","type":"article","venue":"","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"National Research Council Canada; University of Ottawa","funders":"","keywords":"Adversarial system; Hindi; Robustness (evolution); Computer science; German; Training set; Artificial intelligence; Natural language processing; Language model; Test data; Focus (optics); Labeled data; Machine learning; Linguistics","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":[0.002310957,0.0001345569,0.00019072,0.0003422229,0.00009256766,0.00004750895,0.000825449,0.00007676127,0.00009262888],"category_scores_gemma":[0.001300918,0.0001259019,0.00006601456,0.001560197,0.00002766017,0.0003753997,0.00065137,0.0001400097,0.0001476077],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007206728,"about_ca_system_score_gemma":0.0002193097,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00008011309,"about_ca_topic_score_gemma":0.00002856118,"domain_scores_codex":[0.9976926,0.0002428111,0.0003029934,0.0004231464,0.001041793,0.0002966482],"domain_scores_gemma":[0.9986378,0.0001923327,0.0001011836,0.0005474744,0.0004230317,0.00009816901],"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":[0.00001217942,0.0000148421,0.0002587768,0.000006367251,0.00001253507,0.000003867368,0.001209831,0.913506,0.0008076653,0.002521686,0.0004835721,0.0811627],"study_design_scores_gemma":[0.0008239629,0.00004387913,0.002728014,0.00001561742,0.00001629629,0.000001885668,0.00008195497,0.9920341,0.003256812,0.0005153278,0.0003251086,0.0001570767],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.272962,0.000003429016,0.722095,0.000828281,0.001132548,0.0003105711,8.259304e-7,0.0003719578,0.002295354],"genre_scores_gemma":[0.9227021,4.737386e-7,0.07658867,0.0001253466,0.0001771268,0.00002270839,0.000004933157,0.00001289321,0.0003657653],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.64974,"threshold_uncertainty_score":0.5134131,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04854937151821755,"score_gpt":0.3470135699869087,"score_spread":0.2984641984686912,"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."}}