{"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/building-a-regular-decision-boundary-with","title":"Building a Regular Decision Boundary with Deep Networks","arxiv_id":"1703.01775","date":"2017-03-06","proceeding":"CVPR 2017 7","authors":["Edouard Oyallon"],"abstract":"In this work, we build a generic architecture of Convolutional Neural\nNetworks to discover empirical properties of neural networks. Our first\ncontribution is to introduce a state-of-the-art framework that depends upon few\nhyper parameters and to study the network when we vary them. It has no max\npooling, no biases, only 13 layers, is purely convolutional and yields up to\n95.4% and 79.6% accuracy respectively on CIFAR10 and CIFAR100. We show that the\nnonlinearity of a deep network does not need to be continuous, non expansive or\npoint-wise, to achieve good performance. We show that increasing the width of\nour network permits being competitive with very deep networks. Our second\ncontribution is an analysis of the contraction and separation properties of\nthis network. Indeed, a 1-nearest neighbor classifier applied on deep features\nprogressively improves with depth, which indicates that the representation is\nprogressively more regular. Besides, we defined and analyzed local support\nvectors that separate classes locally. All our experiments are reproducible and\ncode is available online, based on TensorFlow.","url_abs":"http://arxiv.org/abs/1703.01775v1","url_pdf":"http://arxiv.org/pdf/1703.01775v1.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":"building-a-regular-decision-boundary-with","repo_url":"https://github.com/edouardoyallon/deep_separation_contraction","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1703.01775","atlas_url":"https://app.syntology.ai/?focus=1703.01775","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}