{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/contrast-to-divide-self-supervised-pre-1","title":"Contrast to Divide: Self-Supervised Pre-Training for Learning with Noisy Labels","arxiv_id":"2103.13646","date":"2021-03-25","proceeding":null,"authors":["Evgenii Zheltonozhskii","Chaim Baskin","Avi Mendelson","Alex M. Bronstein","Or Litany"],"abstract":"The success of learning with noisy labels (LNL) methods relies heavily on the success of a warm-up stage where standard supervised training is performed using the full (noisy) training set. In this paper, we identify a \"warm-up obstacle\": the inability of standard warm-up stages to train high quality feature extractors and avert memorization of noisy labels. We propose \"Contrast to Divide\" (C2D), a simple framework that solves this problem by pre-training the feature extractor in a self-supervised fashion. Using self-supervised pre-training boosts the performance of existing LNL approaches by drastically reducing the warm-up stage's susceptibility to noise level, shortening its duration, and improving extracted feature quality. C2D works out of the box with existing methods and demonstrates markedly improved performance, especially in the high noise regime, where we get a boost of more than 27% for CIFAR-100 with 90% noise over the previous state of the art. In real-life noise settings, C2D trained on mini-WebVision outperforms previous works both in WebVision and ImageNet validation sets by 3% top-1 accuracy. We perform an in-depth analysis of the framework, including investigating the performance of different pre-training approaches and estimating the effective upper bound of the LNL performance with semi-supervised learning. Code for reproducing our experiments is available at https://github.com/ContrastToDivide/C2D","url_abs":"https://arxiv.org/abs/2103.13646v2","url_pdf":"https://arxiv.org/pdf/2103.13646v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"contrast-to-divide-self-supervised-pre-1","repo_url":"https://github.com/ContrastToDivide/C2D","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"learning-with-noisy-labels","task_name":"Learning with noisy labels"},{"task_slug":"memorization","task_name":"Memorization"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/image-classification-on-cifar-10-with-noisy","task":"Image Classification","dataset":"CIFAR-10 (with noisy labels)","model":"C2D (ELR+ with SimCLR, ResNet-34)","rank_in_archive_order":2,"of":4,"metrics":{"Accuracy (under 20% Sym. label noise)":"96.74 ± 0.12","Accuracy (under 50% Sym. label noise)":"95.55 ± 0.32","Accuracy (under 80% Sym. label noise)":"93.11 ± 0.70","Accuracy (under 90% Sym. label noise)":"89.30 ± 0.21","Accuracy (under 95% Sym. label noise)":"80.21 ± 1.91"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-cifar-10-with-noisy","task":"Image Classification","dataset":"CIFAR-10 (with noisy labels)","model":"C2D (DivideMix with SimCLR, ResNet-18)","rank_in_archive_order":4,"of":4,"metrics":{"Accuracy (under 20% Sym. label noise)":"96.23 ± 0.09","Accuracy (under 50% Sym. label noise)":"95.15 ± 0.16","Accuracy (under 80% Sym. label noise)":"94.30 ± 0.12","Accuracy (under 90% Sym. label noise)":"93.42 ± 0.09","Accuracy (under 95% Sym. label noise)":"87.72 ± 2.21"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-clothing1m","task":"Image Classification","dataset":"Clothing1M","model":"ELR+ with C2D (ResNet-50)","rank_in_archive_order":17,"of":51,"metrics":{"Accuracy":"74.58 ± 0.15%"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-mini-webvision-1-0","task":"Image Classification","dataset":"mini WebVision 1.0","model":"DivideMix with C2D (ResNet-50)","rank_in_archive_order":14,"of":47,"metrics":{"ImageNet Top-1 Accuracy":"78.57 ± 0.37","ImageNet Top-5 Accuracy":"93.04 ± 0.10","Top-1 Accuracy":"79.42 ± 0.34","Top-5 Accuracy":"92.32 ± 0.33"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=2103.13646","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2103.13646"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ContrastToDivide/C2D","reach":null}],"summary":{"ran":1,"ran_violates":1,"ran_fixture":2,"ran_honours":1,"ran_draft_wrong":1,"unverified":1},"by_repo_kind":{"official":{"samples":7,"ran":6,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"0f9f54d5a814968a","entry":"CrossEntropyLoss","repo":"ContrastToDivide/C2D","repo_kind":"official","path":"cross_entropy.py","file_url":"https://github.com/ContrastToDivide/C2D/blob/HEAD/cross_entropy.py","link_basis":"first_harvest_node","language":"python","status":"ran","verification_level":1,"contract_check":null,"metamorphic_tier":"invariant","behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"0f9f54d5a814968a"}},{"code_sha256_prefix":"1fa7bb764f450073","entry":"_is_long","repo":"ContrastToDivide/C2D","repo_kind":"official","path":"cross_entropy.py","file_url":"https://github.com/ContrastToDivide/C2D/blob/HEAD/cross_entropy.py","link_basis":"first_harvest_node","language":"python","status":"ran_violates","verification_level":1,"contract_check":"VIOLATES","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"1fa7bb764f450073"}},{"code_sha256_prefix":"a8fa6f956d6f3de1","entry":"cross_entropy","repo":"ContrastToDivide/C2D","repo_kind":"official","path":"cross_entropy.py","file_url":"https://github.com/ContrastToDivide/C2D/blob/HEAD/cross_entropy.py","link_basis":"first_harvest_node","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"a8fa6f956d6f3de1"}},{"code_sha256_prefix":"6944acdaa94d65ef","entry":"linear_rampup","repo":"ContrastToDivide/C2D","repo_kind":"official","path":"main_cifar.py","file_url":"https://github.com/ContrastToDivide/C2D/blob/HEAD/main_cifar.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"6944acdaa94d65ef"}},{"code_sha256_prefix":"cfcd370121d1305f","entry":"onehot","repo":"ContrastToDivide/C2D","repo_kind":"official","path":"cross_entropy.py","file_url":"https://github.com/ContrastToDivide/C2D/blob/HEAD/cross_entropy.py","link_basis":"first_harvest_node","language":"python","status":"ran_draft_wrong","verification_level":1,"contract_check":"MISDECLARED","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"cfcd370121d1305f"}},{"code_sha256_prefix":"46e004362de7a8d3","entry":"smoothing","repo":"ContrastToDivide/C2D","repo_kind":"official","path":"cross_entropy.py","file_url":"https://github.com/ContrastToDivide/C2D/blob/HEAD/cross_entropy.py","link_basis":"first_harvest_node","language":"python","status":"ran_fixture","verification_level":1,"contract_check":"RAISES","metamorphic_tier":null,"behaviour_fingerprint":true,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"46e004362de7a8d3"}},{"code_sha256_prefix":"371eb42215de99b0","entry":"linear_rampup","repo":"ContrastToDivide/C2D","repo_kind":"official","path":"Train_webvision.py","file_url":"https://github.com/ContrastToDivide/C2D/blob/HEAD/Train_webvision.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"371eb42215de99b0"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}