Efficient Backdoor Attacks for Deep Neural Networks in Real-world Scenarios

arXiv:2306.08386v2 Announce Type: replace-cross
Abstract: Recent deep neural networks (DNNs) have came to rely on vast amounts of training data, providing an opportunity for malicious attackers to exploit and contaminate the data to carry out backdoor attacks. However, existing backdoor attack methods make unrealistic assumptions, assuming that all training data comes from a single source and that attackers have full access to the training data. In this paper, we introduce a more realistic attack scenario where victims collect data from …

arxiv attacks backdoor cs.cr cs.cv networks neural networks type world