In the big data era, pre-training large vision transformer (ViT) models on massive datasets has become prevalent for enhanced performance on downstream tasks. Visual prompting (VP), introducing learnable task-specific parameters while freezing the pre-trained backbone, offers an efficient adaptation alternative to full fine-tuning. However, the VP’s potential security risks remain unexplored. The following cybersecurity analysts […]
The post SWARM – Switchable Backdoor Attack Against Pre-trained Models appeared first on GBHackers on Security | #1 Globally Trusted Cyber Security News Platform.