Adversarial Example
Adversarial examples are subtly altered inputs designed to fool machine learning models, primarily deep neural networks (DNNs), into making incorrect predictions. Current research focuses on improving model robustness against these attacks, exploring techniques like ensemble methods, multi-objective representation learning, and adversarial training, often applied to architectures such as ResNets and Vision Transformers. Understanding and mitigating the threat of adversarial examples is crucial for ensuring the reliability and security of AI systems across diverse applications, from image classification and natural language processing to malware detection and autonomous driving. The development of robust defenses and effective attack detection methods remains a significant area of ongoing investigation.
Papers
Mitigating Adversarial Vulnerability through Causal Parameter Estimation by Adversarial Double Machine Learning
Byung-Kwan Lee, Junho Kim, Yong Man Ro
Alleviating the Effect of Data Imbalance on Adversarial Training
Guanlin Li, Guowen Xu, Tianwei Zhang
Vulnerability-Aware Instance Reweighting For Adversarial Training
Olukorede Fakorede, Ashutosh Kumar Nirala, Modeste Atsague, Jin Tian
Does Saliency-Based Training bring Robustness for Deep Neural Networks in Image Classification?
Ali Karkehabadi
Enrollment-stage Backdoor Attacks on Speaker Recognition Systems via Adversarial Ultrasound
Xinfeng Li, Junning Ze, Chen Yan, Yushi Cheng, Xiaoyu Ji, Wenyuan Xu
Boosting Adversarial Transferability with Learnable Patch-wise Masks
Xingxing Wei, Shiji Zhao