Robustness-Congruent Adversarial Training for Secure Machine Learning Model Updates

arXiv:2402.17390v1 Announce Type: cross
Abstract: Machine-learning models demand for periodic updates to improve their average accuracy, exploiting novel architectures and additional data. However, a newly-updated model may commit mistakes that the previous model did not make. Such misclassifications are referred to as negative flips, and experienced by users as a regression of performance. In this work, we show that this problem also affects robustness to adversarial examples, thereby hindering the development of secure model update practices. In particular, when updating …

accuracy adversarial architectures arxiv commit cs.cr cs.lg data demand exploiting machine machine learning may mistakes novel robustness training updates