{"id":"W3132556242","doi":"10.71781/10609","title":"Benchmarking bias mitigation algorithms in representation learning through fairness metrics","year":2022,"lang":"en","type":"article","venue":"Papyrus : Institutional Repository (Université de Montréal)","topic":"Ethics and Social Impacts of AI","field":"Social Sciences","cited_by":10,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal","funders":"McGill University; École Polytechnique Fédérale de Lausanne; Microsoft Research","keywords":"Benchmarking; Computer science; Debiasing; Machine learning; Benchmark (surveying); Artificial intelligence; Deep learning; Artificial neural network; Representation (politics); Deep neural networks; Feature learning; Data mining; Algorithm","routes":{"ca_aff":true,"ca_fund":true,"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":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.02657157,0.001830897,0.001988178,0.002472847,0.001074992,0.004179822,0.003040242,0.003175125,0.002842013],"category_scores_gemma":[0.09195638,0.0005888888,0.001275066,0.001888019,0.002140072,0.004813472,0.003313215,0.003196435,0.0009934561],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003210846,"about_ca_system_score_gemma":0.004902755,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008170178,"about_ca_topic_score_gemma":0.006095551,"domain_scores_codex":[0.9866571,0.00713189,0.0008347234,0.001885028,0.002728726,0.0007625053],"domain_scores_gemma":[0.9478831,0.03761781,0.002243163,0.006093866,0.005268732,0.0008933425],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001216343,0.0003596837,0.01025357,0.0004955535,0.0003629518,0.00007492225,0.0003085841,0.6703907,0.002838938,0.020963,0.002855917,0.2898799],"study_design_scores_gemma":[0.00007942634,0.0002311576,0.001126273,0.00007351489,0.00005467703,0.00004351707,0.00006108699,0.9746372,0.003697847,0.01851183,0.00145642,0.00002694206],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1306008,0.003953989,0.8543003,0.001656277,0.0003174776,0.0003505738,0.0003171538,0.002556636,0.005946876],"genre_scores_gemma":[0.740688,0.0009112214,0.2522627,0.0006117097,0.0001673235,0.0004047578,0.0007681315,0.0004330482,0.003753131],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9734284,"threshold_uncertainty_score":0.1405256,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03166011230794143,"score_gpt":0.2630295987903246,"score_spread":0.2313694864823832,"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."}}