{"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/a-note-on-hyperparameters-in-black-box","title":"A note on hyperparameters in black-box adversarial examples","arxiv_id":"1811.06539","date":"2018-11-15","proceeding":null,"authors":["Jamie Hayes"],"abstract":"Since Biggio et al. (2013) and Szegedy et al. (2013) first drew attention to\nadversarial examples, there has been a flood of research into defending and\nattacking machine learning models. However, almost all proposed attacks assume\nwhite-box access to a model. In other words, the attacker is assumed to have\nperfect knowledge of the models weights and architecture. With this insider\nknowledge, a white-box attack can leverage gradient information to craft\nadversarial examples. Black-box attacks assume no knowledge of the model\nweights or architecture. These attacks craft adversarial examples using\ninformation only contained in the logits or hard classification label. Here, we\nassume the attacker can use the logits in order to find an adversarial example.\nEmpirically, we show that 2-sided stochastic gradient estimation techniques are\nnot sensitive to scaling parameters, and can be used to mount powerful\nblack-box attacks requiring relatively few model queries.","url_abs":"http://arxiv.org/abs/1811.06539v1","url_pdf":"http://arxiv.org/pdf/1811.06539v1.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":"a-note-on-hyperparameters-in-black-box","repo_url":"https://github.com/jhayes14/black-box-attacks","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}