1 paper · 1 filter
Jiale Cai, Gezheng Xu, Zhihao Li +6
Unlearnable examples (UEs) protect training data by injecting imperceptible perturbations so that models fail to extract exploitable representations. In this paper, we reveal that…