Aliasing Backdoor Attacks on Pre-trained ModelsPublished in Proceedings of the 32nd USENIX Security Symposium (USENIX), 2023Share on Twitter Facebook Google+ LinkedIn Previous Next