Leo Yu Zhang

Self-Supervised Adversarial Example Detection by Disentangled Representation

2022

Deep learning models are known to be vulnerable to adversarial examples that are elaborately designed for malicious purposes and are imperceptible to the human perceptual system. Autoencoder, when trained solely over benign examples, has been …

Backdoor Attack on Deep Neural Networks in Perception Domain

2023

As deep neural networks (DNNs) are widely deployed in various applications, the security of pretrained DNNs is crucial since backdoors can be introduced through poisoned training. A backdoored DNN model works properly when benign inputs …

WiP: Towards Zero Trust Authentication in Critical Industrial Infrastructures with PRISM

2023 · Lecture notes in computer science

With the increasing threat of cyber attacks on critical infrastructures, the need for robust security measures has become more pressing. In response, decentralized secure computation has gained traction as an effective approach to minimizing the …

PB-UAP: Hybrid Universal Adversarial Attack For Image Segmentation

2024 · arXiv (Cornell University)

With the rapid advancement of deep learning, the model robustness has become a significant research hotspot, \ie, adversarial attacks on deep neural networks. Existing works primarily focus on image classification tasks, aiming to alter the …

Data Duplication: A Novel Multi-Purpose Attack Paradigm in Machine Unlearning

2025 · arXiv (Cornell University)

Duplication is a prevalent issue within datasets. Existing research has demonstrated that the presence of duplicated data in training datasets can significantly influence both model performance and data privacy. However, the impact of data duplication …