Yaopeng Wang’s research while affiliated with Fuzhou University and other places

High cost of training time caused by multi-step adversarial example generation is a major challenge in adversarial training. Previous methods try to reduce the computational burden of adversarial training using single-step adversarial example generation schemes, which can effectively improve the efficiency but also introduce the problem of “catastrophic overfitting”, where the robust accuracy against Fast Gradient Sign Method (FGSM) can achieve nearby 100% whereas the robust accuracy against Projected Gradient Descent (PGD) suddenly drops to 0% over a single epoch. To address this issue, we focus on single-step adversarial training scheme in this paper and propose a novel Fast Gradient Sign Method with PGD Regularization (FGSMPR) to boost the efficiency of adversarial training without catastrophic overfitting. Our core observation is that single-step adversarial training can not simultaneously learn robust internal representations of FGSM and PGD adversarial examples. Therefore, we design a PGD regularization term to encourage similar embeddings of FGSM and PGD adversarial examples. The experiments demonstrate that our proposed method can train a robust deep network for L∞-perturbations with FGSM adversarial training and reduce the gap to multi-step adversarial training.

High cost of training time caused by multi-step adversarial example generation is a major challenge in adversarial training. Previous methods try to reduce the computational burden of adversarial training using single-step adversarial example generation schemes, which can effectively improve the efficiency but also introduce the problem of catastrophic overfitting, where the robust accuracy against Fast Gradient Sign Method (FGSM) can achieve nearby 100\% whereas the robust accuracy against Projected Gradient Descent (PGD) suddenly drops to 0\% over a single epoch. To address this problem, we propose a novel Fast Gradient Sign Method with PGD Regularization (FGSMPR) to boost the efficiency of adversarial training without catastrophic overfitting. Our core idea is that single-step adversarial training can not learn robust internal representations of FGSM and PGD adversarial examples. Therefore, we design a PGD regularization term to encourage similar embeddings of FGSM and PGD adversarial examples. The experiments demonstrate that our proposed method can train a robust deep network for L$_\infty$-perturbations with FGSM adversarial training and reduce the gap to multi-step adversarial training.

Deep Learning (DL) algorithms based on artificial neural networks have achieved remarkable success and are being extensively applied in a variety of application domains, ranging from image classification, automatic driving, natural language processing to medical diagnosis, credit risk assessment, intrusion detection. However, the privacy and security issues of DL have been revealed that the DL model can be stolen or reverse engineered, sensitive training data can be inferred, even a recognizable face image of the victim can be recovered. Besides, the recent works have found that the DL model is vulnerable to adversarial examples perturbed by imperceptible noised, which can lead the DL model to predict wrongly with high confidence. In this paper, we first briefly introduces the four types of attacks and privacy-preserving techniques in DL. We then review and summarize the attack and defense methods associated with DL privacy and security in recent years. To demonstrate that security threats really exist in the real world, we also reviewed the adversarial attacks under the physical condition. Finally, we discuss current challenges and open problems regarding privacy and security issues in DL.