Backdoor Defense via Suppressing Model Shortcuts
This addresses a security vulnerability in DNNs for applications requiring robust models, but it is incremental as it builds on existing understanding of skip connections and backdoor mechanisms.
The paper tackles backdoor attacks in deep neural networks by suppressing skip connections to reduce the attack success rate, achieving significant decreases in ASR while maintaining high benign accuracy through fine-tuning.
Recent studies have demonstrated that deep neural networks (DNNs) are vulnerable to backdoor attacks during the training process. Specifically, the adversaries intend to embed hidden backdoors in DNNs so that malicious model predictions can be activated through pre-defined trigger patterns. In this paper, we explore the backdoor mechanism from the angle of the model structure. We select the skip connection for discussions, inspired by the understanding that it helps the learning of model `shortcuts' where backdoor triggers are usually easier to be learned. Specifically, we demonstrate that the attack success rate (ASR) decreases significantly when reducing the outputs of some key skip connections. Based on this observation, we design a simple yet effective backdoor removal method by suppressing the skip connections in critical layers selected by our method. We also implement fine-tuning on these layers to recover high benign accuracy and to further reduce ASR. Extensive experiments on benchmark datasets verify the effectiveness of our method.