{"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/detecting-adversarial-samples-from-artifacts","title":"Detecting Adversarial Samples from Artifacts","arxiv_id":"1703.00410","date":"2017-03-01","proceeding":null,"authors":["Reuben Feinman","Ryan R. Curtin","Saurabh Shintre","Andrew B. Gardner"],"abstract":"Deep neural networks (DNNs) are powerful nonlinear architectures that are\nknown to be robust to random perturbations of the input. However, these models\nare vulnerable to adversarial perturbations--small input changes crafted\nexplicitly to fool the model. In this paper, we ask whether a DNN can\ndistinguish adversarial samples from their normal and noisy counterparts. We\ninvestigate model confidence on adversarial samples by looking at Bayesian\nuncertainty estimates, available in dropout neural networks, and by performing\ndensity estimation in the subspace of deep features learned by the model. The\nresult is a method for implicit adversarial detection that is oblivious to the\nattack algorithm. We evaluate this method on a variety of standard datasets\nincluding MNIST and CIFAR-10 and show that it generalizes well across different\narchitectures and attacks. Our findings report that 85-93% ROC-AUC can be\nachieved on a number of standard classification tasks with a negative class\nthat consists of both normal and noisy samples.","url_abs":"http://arxiv.org/abs/1703.00410v3","url_pdf":"http://arxiv.org/pdf/1703.00410v3.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":"detecting-adversarial-samples-from-artifacts","repo_url":"https://github.com/rfeinman/detecting-adversarial-samples","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"detecting-adversarial-samples-from-artifacts","repo_url":"https://github.com/300wonplz/Detecting-Adversarial-Samples-from-Artifacts-pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"detecting-adversarial-samples-from-artifacts","repo_url":"https://github.com/Hyeongmin-Cho/Detecting-Adversarial-Samples-from-Artifacts-pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"density-estimation","task_name":"Density Estimation"}],"methods":[{"method_slug":"dropout","method_name":"Dropout"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1703.00410","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}