{"id":"W2962762541","doi":"","title":"TRAINING DEEP NEURAL NETWORKS ON NOISY LABELS WITH BOOTSTRAPPING","year":2015,"lang":"en","type":"article","venue":"arXiv (Cornell University)","topic":"Face recognition and analysis","field":"Computer Science","cited_by":330,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Computer science; Overfitting; Artificial intelligence; MNIST database; Robustness (evolution); Machine learning; Pattern recognition (psychology); Deep learning; Deep neural networks; Dropout (neural networks); Bootstrapping (finance); Scalability; Consistency (knowledge bases); Facial recognition system; Artificial neural network","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1363999653471787,"score_gpt":0.1858086680080738,"score_spread":0.04940870266089506,"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."},"direct_labels":[],"classifier":{"requested":true,"available":true,"version":"metacan-v1-d91a1de5be90","frame_rows_covered":4299418,"decision_targets":["metaresearch","metaepi_narrow","metaepi_broad","bibliometrics","sts","scholarly_communication","open_science","research_integrity","randomized_trial","nonrandomized_trial","observational","systematic_review","meta_analysis","case_report","qualitative","simulation_or_modeling","bench_or_experimental","theoretical_or_conceptual","not_applicable"],"score_encoding":"uint16_le_65535","score_resolution":0.000015259021896696422,"interpretation":"Scores imitate each teacher on the enriched screening sample; they are not calibrated prevalence probabilities for the full frame.","warning":null},"prediction":{"classifier_version":"metacan-v1-d91a1de5be90","candidate_union":["simulation_or_modeling","theoretical_or_conceptual"],"consensus_intersection":["simulation_or_modeling"],"score_encoding":"uint16_le_65535","score_resolution":0.000015259021896696422,"scores":{"codex":{"metaresearch":0.0013885709925993744,"metaepi_narrow":0,"metaepi_broad":0.000015259021896696422,"bibliometrics":0.004257267109178301,"sts":0.000015259021896696422,"scholarly_communication":0.00006103608758678569,"open_science":0.00024414435034714275,"research_integrity":0.00009155413138017853,"randomized_trial":0.00009155413138017853,"nonrandomized_trial":0.000030518043793392844,"observational":0.00022888532845044633,"systematic_review":0,"meta_analysis":0.000030518043793392844,"case_report":0.00006103608758678569,"qualitative":0.00015259021896696422,"simulation_or_modeling":0.8144045166704814,"bench_or_experimental":0.00006103608758678569,"theoretical_or_conceptual":0.04924086366063935,"not_applicable":0.00007629510948348211,"design_other":0.314702067597467},"gemma":{"metaresearch":0.0022583352407110706,"metaepi_narrow":0.000015259021896696422,"metaepi_broad":0.000015259021896696422,"bibliometrics":0.00392156862745098,"sts":0.000015259021896696422,"scholarly_communication":0.00007629510948348211,"open_science":0.000030518043793392844,"research_integrity":0.000045777065690089265,"randomized_trial":0.00006103608758678569,"nonrandomized_trial":0.00006103608758678569,"observational":0.005462729839017319,"systematic_review":0.00012207217517357137,"meta_analysis":0.000030518043793392844,"case_report":0.0001373311970702678,"qualitative":0.0004119935912108034,"simulation_or_modeling":0.9744106202792401,"bench_or_experimental":0.000015259021896696422,"theoretical_or_conceptual":0.005554283970397498,"not_applicable":0.0006561379415579462}},"interpretation":"Scores imitate each teacher on the enriched screening sample; they are not calibrated prevalence probabilities for the full frame."}}