{"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/mixup-as-locally-linear-out-of-manifold","title":"MixUp as Locally Linear Out-Of-Manifold Regularization","arxiv_id":"1809.02499","date":"2018-09-07","proceeding":null,"authors":["Hongyu Guo","Yongyi Mao","Richong Zhang"],"abstract":"MixUp is a recently proposed data-augmentation scheme, which linearly\ninterpolates a random pair of training examples and correspondingly the one-hot\nrepresentations of their labels. Training deep neural networks with such\nadditional data is shown capable of significantly improving the predictive\naccuracy of the current art. The power of MixUp, however, is primarily\nestablished empirically and its working and effectiveness have not been\nexplained in any depth. In this paper, we develop an understanding for MixUp as\na form of \"out-of-manifold regularization\", which imposes certain \"local\nlinearity\" constraints on the model's input space beyond the data manifold.\nThis analysis enables us to identify a limitation of MixUp, which we call\n\"manifold intrusion\". In a nutshell, manifold intrusion in MixUp is a form of\nunder-fitting resulting from conflicts between the synthetic labels of the\nmixed-up examples and the labels of original training data. Such a phenomenon\nusually happens when the parameters controlling the generation of mixing\npolicies are not sufficiently fine-tuned on the training data. To address this\nissue, we propose a novel adaptive version of MixUp, where the mixing policies\nare automatically learned from the data using an additional network and\nobjective function designed to avoid manifold intrusion. The proposed\nregularizer, AdaMixUp, is empirically evaluated on several benchmark datasets.\nExtensive experiments demonstrate that AdaMixUp improves upon MixUp when\napplied to the current art of deep classification models.","url_abs":"http://arxiv.org/abs/1809.02499v3","url_pdf":"http://arxiv.org/pdf/1809.02499v3.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":"mixup-as-locally-linear-out-of-manifold","repo_url":"https://github.com/SITE5039/AdaMixUp","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"mixup-as-locally-linear-out-of-manifold","repo_url":"https://github.com/makeyourownmaker/mixup","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"GPL-2.0"}}],"tasks":[{"task_slug":"data-augmentation","task_name":"Data Augmentation"}],"methods":[{"method_slug":"mixup","method_name":"Mixup"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1809.02499","atlas_url":"https://app.syntology.ai/?focus=1809.02499","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1809.02499"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-25T09:33:49+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/SITE5039/AdaMixUp","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/makeyourownmaker/mixup","reach":{"status":"ok","spdx":"GPL-2.0"}}],"summary":{"ran_honours":2},"by_repo_kind":{"listed":{"samples":2,"ran":2,"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":2,"samples":[{"code_sha256_prefix":"6da4c3aa8c6bb48b","entry":"extract_images","repo":"SITE5039/AdaMixUp","repo_kind":"listed","path":"GitHub.AdaMixup.AAAI.MNIST.Cifar.py","file_url":"https://github.com/SITE5039/AdaMixUp/blob/HEAD/GitHub.AdaMixup.AAAI.MNIST.Cifar.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":"well_formed","behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"6da4c3aa8c6bb48b"}},{"code_sha256_prefix":"6258fc666d11d4ba","entry":"extract_labels","repo":"SITE5039/AdaMixUp","repo_kind":"listed","path":"GitHub.AdaMixup.AAAI.MNIST.Cifar.py","file_url":"https://github.com/SITE5039/AdaMixUp/blob/HEAD/GitHub.AdaMixup.AAAI.MNIST.Cifar.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"NONE","inline_ok":false,"mcp_get_code":{"code_sha256":"6258fc666d11d4ba"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}