Adversarial Feature Map Pruning for Backdoor

arXiv:2307.11565v2 Announce Type: replace
Abstract: Deep neural networks have been widely used in many critical applications, such as autonomous vehicles and medical diagnosis. However, their security is threatened by backdoor attacks, which are achieved by adding artificial patterns to specific training data. Existing defense strategies primarily focus on using reverse engineering to reproduce the backdoor trigger generated by attackers and subsequently repair the DNN model by adding the trigger into inputs and fine-tuning the model with ground-truth labels. However, once …

abstract adversarial applications artificial arxiv attacks autonomous autonomous vehicles backdoor cs.lg cs.se data defense diagnosis engineering feature focus map medical networks neural networks patterns pruning security strategies training training data type vehicles