LTD: Low Temperature Distillation for Robust Adversarial Training
share
Adversarial training has been widely used to enhance the robustness of the neural network models against adversarial attacks. However, there still a notable gap between the nature accuracy and the robust accuracy. We found one of the reasons is the commonly used labels, one-hot vectors, hinder the learning process for image recognition. In this paper, we proposed a method, called Low Temperature Distillation (LTD), which is based on the knowledge distillation framework to generate the desired soft labels. Unlike the previous work, LTD uses relatively low temperature in the teacher model, and employs different, but fixed, temperatures for the teacher model and the student model. Moreover, we have investigated the methods to synergize the use of nature data and adversarial ones in LTD. Experimental results show that without extra unlabeled data, the proposed method combined with the previous work can achieve 57.72% and 30.36% robust accuracy on CIFAR-10 and CIFAR-100 dataset respectively, which is about 1.21% improvement of the state-of-the-art methods in average.
READ FULL TEXT