Safety and Performance, Why Not Both? Bi-Objective Optimized Model Compression against Heterogeneous Attacks Toward AI Software Deployment

1 code implementation • 2 Jan 2024 • Jie Zhu, Leye Wang, Xiao Han, Anmin Liu, Tao Xie

To mitigate this issue, AI software compression plays a crucial role, which aims to compress model size while keeping high performance.

Inference Attack Membership Inference Attack +1

Sampling-based Fast Gradient Rescaling Method for Highly Transferable Adversarial Attacks

1 code implementation • 6 Jul 2023 • Xu Han, Anmin Liu, Chenxuan Yao, Yanbo Fan, Kun He

In either case, the common gradient-based methods generally use the sign function to generate perturbations on the gradient update, that offers a roughly correct direction and has gained great success.

Sampling-based Fast Gradient Rescaling Method for Highly Transferable Adversarial Attacks

1 code implementation • 6 Apr 2022 • Xu Han, Anmin Liu, Yifeng Xiong, Yanbo Fan, Kun He

Deviation between the original gradient and the generated noises may lead to inaccurate gradient update estimation and suboptimal solutions for adversarial transferability, which is crucial for black-box attacks.