Anmin Liu

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

2022 · arXiv (Cornell University)

Deep neural networks have shown to be very vulnerable to adversarial examples crafted by adding human-imperceptible perturbations to benign inputs. After achieving impressive attack success rates in the white-box setting, more focus is shifted to …

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

2024 · IEEE Transactions on Software Engineering

The size of deep learning models in artificial intelligence (AI) software is increasing rapidly, which hinders the large-scale deployment on resource-restricted devices (e.g., smartphones). To mitigate this issue, AI software compression plays a crucial role, …