{"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/exploring-the-space-of-adversarial-images","title":"Exploring the Space of Adversarial Images","arxiv_id":"1510.05328","date":"2015-10-19","proceeding":null,"authors":["Pedro Tabacof","Eduardo Valle"],"abstract":"Adversarial examples have raised questions regarding the robustness and\nsecurity of deep neural networks. In this work we formalize the problem of\nadversarial images given a pretrained classifier, showing that even in the\nlinear case the resulting optimization problem is nonconvex. We generate\nadversarial images using shallow and deep classifiers on the MNIST and ImageNet\ndatasets. We probe the pixel space of adversarial images using noise of varying\nintensity and distribution. We bring novel visualizations that showcase the\nphenomenon and its high variability. We show that adversarial images appear in\nlarge regions in the pixel space, but that, for the same task, a shallow\nclassifier seems more robust to adversarial images than a deep convolutional\nnetwork.","url_abs":"http://arxiv.org/abs/1510.05328v5","url_pdf":"http://arxiv.org/pdf/1510.05328v5.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":"exploring-the-space-of-adversarial-images","repo_url":"https://github.com/tabacof/adversarial","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"torch","reach":{"status":"ok","spdx":"GPL-3.0"}},{"paper_slug":"exploring-the-space-of-adversarial-images","repo_url":"https://github.com/luizgh/adversarial","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"torch","reach":{"status":"ok","spdx":"GPL-3.0"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1510.05328","atlas_url":"https://app.syntology.ai/?focus=1510.05328","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}