{"url":"/sota/learning-with-noisy-labels-on-cifar-10n","task":{"name":"Learning with noisy labels","url":"/task/learning-with-noisy-labels","note":null},"dataset":{"name":"CIFAR-10N-Aggregate","url":"/dataset/cifar-10n"},"category":"Computer Vision","categories":["Adversarial","Computer Vision","Medical","Natural Language Processing"],"category_note":null,"description":"Learning with noisy labels means When we say \"noisy labels,\" we mean that an adversary has intentionally messed up the labels, which would have come from a \"clean\" distribution otherwise. This setting can also be used to cast learning from only positive and unlabeled data.","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["Accuracy (mean)"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Accuracy (mean)":"higher"}},"counts":{"rows":26,"rows_with_code":24,"rows_with_paper_page":25,"rows_dated":25,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"ProMix","metrics":{"Accuracy (mean)":"97.39"},"uses_additional_data":false,"paper_date":"2022-07-21","paper":"/paper/promix-combating-label-noise-via-maximizing","paper_url":"https://arxiv.org/abs/2207.10276v4","paper_title":"ProMix: Combating Label Noise via Maximizing Clean Sample Utility","code":"https://github.com/justherozen/promix","n_code_links":1,"syntology":{"n_ran":0,"n_unverified":1,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":2,"model":"PSSCL","metrics":{"Accuracy (mean)":"96.41"},"uses_additional_data":false,"paper_date":"2024-12-18","paper":"/paper/psscl-a-progressive-sample-selection","paper_url":"https://www.sciencedirect.com/science/article/abs/pii/S0031320324010355","paper_title":"PSSCL: A progressive sample selection framework with contrastive loss designed for noisy labels","code":"https://github.com/LanXiaoPang613/PSSCL","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"PGDF","metrics":{"Accuracy (mean)":"96.11"},"uses_additional_data":false,"paper_date":"2021-12-02","paper":"/paper/sample-prior-guided-robust-model-learning-to","paper_url":"https://arxiv.org/abs/2112.01197v3","paper_title":"Sample Prior Guided Robust Model Learning to Suppress Noisy Labels","code":"https://github.com/bupt-ai-cz/PGDF","n_code_links":1,"syntology":null},{"rank_in_archive_order":4,"model":"SOP+","metrics":{"Accuracy (mean)":"95.61"},"uses_additional_data":false,"paper_date":"2022-02-28","paper":"/paper/robust-training-under-label-noise-by-over","paper_url":"https://arxiv.org/abs/2202.14026v2","paper_title":"Robust Training under Label Noise by Over-parameterization","code":"https://github.com/shengliu66/sop","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":7,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":5,"model":"ILL","metrics":{"Accuracy (mean)":"95.47"},"uses_additional_data":false,"paper_date":"2023-05-22","paper":"/paper/imprecise-label-learning-a-unified-framework","paper_url":"https://arxiv.org/abs/2305.12715v4","paper_title":"Imprecise Label Learning: A Unified Framework for Learning with Various Imprecise Label Configurations","code":"https://github.com/hhhhhhao/general-framework-weak-supervision","n_code_links":1,"syntology":null},{"rank_in_archive_order":6,"model":"CORES*","metrics":{"Accuracy (mean)":"95.25"},"uses_additional_data":false,"paper_date":"2020-10-05","paper":"/paper/learning-with-instance-dependent-label-noise-1","paper_url":"https://arxiv.org/abs/2010.02347v2","paper_title":"Learning with Instance-Dependent Label Noise: A Sample Sieve Approach","code":"https://github.com/UCSC-REAL/cores","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":2,"n_samples":4,"n_pointer_only_licence":4}},{"rank_in_archive_order":7,"model":"Divide-Mix","metrics":{"Accuracy (mean)":"95.01"},"uses_additional_data":false,"paper_date":"2020-02-18","paper":"/paper/dividemix-learning-with-noisy-labels-as-semi-1","paper_url":"https://arxiv.org/abs/2002.07394v1","paper_title":"DivideMix: Learning with Noisy Labels as Semi-supervised Learning","code":"https://github.com/LiJunnan1992/DivideMix","n_code_links":2,"syntology":{"n_ran":0,"n_unverified":2,"n_samples":2,"n_pointer_only_licence":2}},{"rank_in_archive_order":8,"model":"ELR+","metrics":{"Accuracy (mean)":"94.83"},"uses_additional_data":false,"paper_date":"2020-06-30","paper":"/paper/early-learning-regularization-prevents","paper_url":"https://arxiv.org/abs/2007.00151v2","paper_title":"Early-Learning Regularization Prevents Memorization of Noisy Labels","code":"https://github.com/shengliu66/ELR","n_code_links":2,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":9,"model":"PES (Semi)","metrics":{"Accuracy (mean)":"94.66"},"uses_additional_data":false,"paper_date":"2021-06-30","paper":"/paper/understanding-and-improving-early-stopping","paper_url":"https://arxiv.org/abs/2106.15853v1","paper_title":"Understanding and Improving Early Stopping for Learning with Noisy Labels","code":"https://github.com/tmllab/PES","n_code_links":1,"syntology":null},{"rank_in_archive_order":10,"model":"GNL","metrics":{"Accuracy (mean)":"92.57"},"uses_additional_data":false,"paper_date":"2023-08-02","paper":"/paper/generative-noisy-label-learning-by-implicit","paper_url":"https://arxiv.org/abs/2308.01184v2","paper_title":"Partial Label Supervision for Agnostic Generative Noisy Label Learning","code":"https://github.com/lfb-1/gnl","n_code_links":1,"syntology":null},{"rank_in_archive_order":11,"model":"ELR","metrics":{"Accuracy (mean)":"92.38"},"uses_additional_data":false,"paper_date":"2020-06-30","paper":"/paper/early-learning-regularization-prevents","paper_url":"https://arxiv.org/abs/2007.00151v2","paper_title":"Early-Learning Regularization Prevents Memorization of Noisy Labels","code":"https://github.com/shengliu66/ELR","n_code_links":2,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":12,"model":"CAL","metrics":{"Accuracy (mean)":"91.97"},"uses_additional_data":false,"paper_date":"2021-02-10","paper":"/paper/clusterability-as-an-alternative-to-anchor","paper_url":"https://arxiv.org/abs/2102.05291v2","paper_title":"Clusterability as an Alternative to Anchor Points When Learning with Noisy Labels","code":"https://github.com/UCSC-REAL/HOC","n_code_links":2,"syntology":{"n_ran":0,"n_unverified":9,"n_samples":9,"n_pointer_only_licence":0}},{"rank_in_archive_order":13,"model":"Negative-LS","metrics":{"Accuracy (mean)":"91.97"},"uses_additional_data":false,"paper_date":"2021-06-08","paper":"/paper/understanding-generalized-label-smoothing","paper_url":"https://arxiv.org/abs/2106.04149v6","paper_title":"To Smooth or Not? When Label Smoothing Meets Noisy Labels","code":"https://github.com/ucsc-real/negative-label-smoothing","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":14,"model":"F-div","metrics":{"Accuracy (mean)":"91.64"},"uses_additional_data":false,"paper_date":"2020-11-07","paper":"/paper/when-optimizing-f-divergence-is-robust-with-1","paper_url":"https://arxiv.org/abs/2011.03687v3","paper_title":"When Optimizing $f$-divergence is Robust with Label Noise","code":"https://github.com/weijiaheng/Robust-f-divergence-measures","n_code_links":2,"syntology":null},{"rank_in_archive_order":15,"model":"Positive-LS","metrics":{"Accuracy (mean)":"91.57"},"uses_additional_data":false,"paper_date":"2020-03-05","paper":"/paper/does-label-smoothing-mitigate-label-noise","paper_url":"https://arxiv.org/abs/2003.02819v1","paper_title":"Does label smoothing mitigate label noise?","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":16,"model":"JoCoR","metrics":{"Accuracy (mean)":"91.44"},"uses_additional_data":false,"paper_date":"2020-03-05","paper":"/paper/combating-noisy-labels-by-agreement-a-joint","paper_url":"https://arxiv.org/abs/2003.02752v3","paper_title":"Combating noisy labels by agreement: A joint training method with co-regularization","code":"https://github.com/hongxin001/JoCoR","n_code_links":2,"syntology":null},{"rank_in_archive_order":17,"model":"CORES","metrics":{"Accuracy (mean)":"91.23"},"uses_additional_data":false,"paper_date":"2020-10-05","paper":"/paper/learning-with-instance-dependent-label-noise-1","paper_url":"https://arxiv.org/abs/2010.02347v2","paper_title":"Learning with Instance-Dependent Label Noise: A Sample Sieve Approach","code":"https://github.com/UCSC-REAL/cores","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":2,"n_samples":4,"n_pointer_only_licence":4}},{"rank_in_archive_order":18,"model":"Co-Teaching","metrics":{"Accuracy (mean)":"91.20"},"uses_additional_data":false,"paper_date":"2018-04-18","paper":"/paper/co-teaching-robust-training-of-deep-neural","paper_url":"http://arxiv.org/abs/1804.06872v3","paper_title":"Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labels","code":"https://github.com/bhanML/Co-teaching","n_code_links":5,"syntology":{"n_ran":7,"n_unverified":0,"n_samples":7,"n_pointer_only_licence":7}},{"rank_in_archive_order":19,"model":"Peer Loss","metrics":{"Accuracy (mean)":"90.75"},"uses_additional_data":false,"paper_date":"2019-10-08","paper":"/paper/peer-loss-functions-learning-from-noisy","paper_url":"https://arxiv.org/abs/1910.03231v7","paper_title":"Peer Loss Functions: Learning from Noisy Labels without Knowing Noise Rates","code":"https://github.com/weijiaheng/Multi-class-Peer-Loss-functions","n_code_links":2,"syntology":{"n_ran":4,"n_unverified":2,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":20,"model":"Co-Teaching+","metrics":{"Accuracy (mean)":"90.61"},"uses_additional_data":false,"paper_date":"2019-01-14","paper":"/paper/how-does-disagreement-help-generalization","paper_url":"https://arxiv.org/abs/1901.04215v3","paper_title":"How does Disagreement Help Generalization against Label Corruption?","code":"https://github.com/xingruiyu/coteaching_plus","n_code_links":3,"syntology":null},{"rank_in_archive_order":21,"model":"VolMinNet","metrics":{"Accuracy (mean)":"89.70"},"uses_additional_data":false,"paper_date":"2021-02-04","paper":"/paper/provably-end-to-end-label-noise-learning","paper_url":"https://arxiv.org/abs/2102.02400v4","paper_title":"Provably End-to-end Label-Noise Learning without Anchor Points","code":"https://github.com/xuefeng-li1/Provably-end-to-end-label-noise-learning-without-anchor-points","n_code_links":1,"syntology":null},{"rank_in_archive_order":22,"model":"T-Revision","metrics":{"Accuracy (mean)":"88.52"},"uses_additional_data":false,"paper_date":"2019-06-01","paper":"/paper/190600189","paper_url":"https://arxiv.org/abs/1906.00189v2","paper_title":"Are Anchor Points Really Indispensable in Label-Noise Learning?","code":"https://github.com/xiaoboxia/T-Revision","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":23,"model":"Forward-T","metrics":{"Accuracy (mean)":"88.24"},"uses_additional_data":false,"paper_date":"2016-09-13","paper":"/paper/making-deep-neural-networks-robust-to-label","paper_url":"http://arxiv.org/abs/1609.03683v2","paper_title":"Making Deep Neural Networks Robust to Label Noise: a Loss Correction Approach","code":"https://github.com/yikun2019/PENCIL","n_code_links":2,"syntology":{"n_ran":2,"n_unverified":0,"n_samples":2,"n_pointer_only_licence":2}},{"rank_in_archive_order":24,"model":"Backward-T","metrics":{"Accuracy (mean)":"88.13"},"uses_additional_data":false,"paper_date":"2016-09-13","paper":"/paper/making-deep-neural-networks-robust-to-label","paper_url":"http://arxiv.org/abs/1609.03683v2","paper_title":"Making Deep Neural Networks Robust to Label Noise: a Loss Correction Approach","code":"https://github.com/yikun2019/PENCIL","n_code_links":2,"syntology":{"n_ran":2,"n_unverified":0,"n_samples":2,"n_pointer_only_licence":2}},{"rank_in_archive_order":25,"model":"GCE","metrics":{"Accuracy (mean)":"87.85"},"uses_additional_data":false,"paper_date":"2018-05-20","paper":"/paper/generalized-cross-entropy-loss-for-training","paper_url":"http://arxiv.org/abs/1805.07836v4","paper_title":"Generalized Cross Entropy Loss for Training Deep Neural Networks with Noisy Labels","code":"https://github.com/AlanChou/Truncated-Loss","n_code_links":4,"syntology":null},{"rank_in_archive_order":26,"model":"CE","metrics":{"Accuracy (mean)":"87.77"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 6,795 of the 9,581 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9581,"papers_checked":6795,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":2785},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-24T18:15:14+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":14,"rows_with_any_sample_ran":11,"distinct_papers_with_graph_line":11,"distinct_papers_with_any_sample_ran":8,"samples_over_distinct_papers":{"n_ran":23,"n_unverified":23,"n_samples":46,"n_pointer_only_licence":21,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":28,"n_unverified":25,"n_samples":53,"n_pointer_only_licence":27,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}