Group-based Robustness: A General Framework for Customized Robustness in the Real World

arXiv:2306.16614v3 Announce Type: replace-cross
Abstract: Machine-learning models are known to be vulnerable to evasion attacks that perturb model inputs to induce misclassifications. In this work, we identify real-world scenarios where the true threat cannot be assessed accurately by existing attacks. Specifically, we find that conventional metrics measuring targeted and untargeted robustness do not appropriately reflect a model's ability to withstand attacks from one set of source classes to another set of target classes. To address the shortcomings of existing methods, …

arxiv attacks cs.ai cs.cr cs.cv cs.lg evasion evasion attacks find framework general identify inputs machine measuring metrics real robustness threat vulnerable work world