{"id":"W2947280697","doi":"10.48550/arxiv.1905.01347","title":"Auditing ImageNet: Towards a Model-driven Framework for Annotating Demographic Attributes of Large-Scale Image Datasets","year":2019,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Face recognition and analysis","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Annotation; Scale (ratio); Audit; Demographics; Artificial intelligence; Machine learning; Code (set theory); Point (geometry); Variety (cybernetics); Data science; Set (abstract data type); Cartography","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":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0004186858,0.0003428365,0.0005882417,0.000448815,0.0001957817,0.0001360114,0.001644089,0.0003100608,0.00002090604],"category_scores_gemma":[0.0001428308,0.0003999976,0.0005753515,0.00091353,0.00009808794,0.000492441,0.001759688,0.0005007336,0.00002481991],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007514464,"about_ca_system_score_gemma":0.0002126129,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004847947,"about_ca_topic_score_gemma":0.00004501145,"domain_scores_codex":[0.9977075,0.0001156232,0.0003602657,0.001171438,0.0001489313,0.0004962439],"domain_scores_gemma":[0.9973772,0.0002436959,0.0006400194,0.001170005,0.0004148227,0.0001542374],"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":[0.00003250966,0.0002678857,0.004618474,0.0007695374,0.0004957527,0.00004806448,0.0005635894,0.9612445,0.0007163876,0.02784798,0.001066661,0.002328658],"study_design_scores_gemma":[0.0004266862,0.00003037242,0.0002508515,0.0002856189,0.0001795022,0.000001066434,0.0001919981,0.9651849,0.0008159691,0.03215572,0.00006255077,0.000414727],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.08808322,0.00005268205,0.9080356,0.0001332573,0.0001449852,0.0003606484,0.002937153,0.0001290017,0.0001234896],"genre_scores_gemma":[0.748413,0.0001222953,0.2506233,0.00009902628,0.00003805471,0.000002875547,0.0006181653,0.00002012059,0.00006319248],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.6603298,"threshold_uncertainty_score":0.9998452,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0642147363907538,"score_gpt":0.2339749395754144,"score_spread":0.1697602031846606,"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."}}