A Neural Rejection System Against Universal Adversarial Perturbations in Radio Signal Classification

2025/06/13 by Lu Zhang, Zhang, Lu, Sangarapillai Lambotharan +5
Computer Science · Engineering · #Advanced SAR Imaging Techniques #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Radar Systems and Signal Processing #Wireless Signal Modulation Classification

paper · pdf · doi:10.48550/arxiv.2506.11901

openalex publication_date 2025/06/13 · openalex created_date 2025/10/11 · openalex updated_date 2026/07/28

Abstract

Advantages of deep learning over traditional methods have been demonstrated for radio signal classification in the recent years. However, various researchers have discovered that even a small but intentional feature perturbation known as adversarial examples can significantly deteriorate the performance of the deep learning based radio signal classification. Among various kinds of adversarial examples, universal adversarial perturbation has gained considerable attention due to its feature of being data independent, hence as a practical strategy to fool the radio signal classification with a high success rate. Therefore, in this paper, we investigate a defense system called neural rejection system to propose against universal adversarial perturbations, and evaluate its performance by generating white-box universal adversarial perturbations. We show that the proposed neural rejection system is able to defend universal adversarial perturbations with significantly higher accuracy than the undefended deep neural network.

Citations

• Deep Neural Rejection against Adversarial Examples
• Classification of Radio Signals and HF Transmission Modes with Deep Learning
• Adversarial Attacks on Deep-Learning Based Radio Signal Classification
• Adversarial Examples: Attacks and Defenses for Deep Learning
• YOLO9000: Better, Faster, Stronger
• Universal adversarial perturbations
• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal\n Networks
• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
• The IBM 2015 English Conversational Telephone Speech Recognition System
• Sequence to Sequence Learning with Neural Networks
• Very Deep Convolutional Networks for Large-Scale Image Recognition
• ImageNet classification with deep convolutional neural networks

Related

• Countermeasures Against Adversarial Examples in Radio Signal Classification
• Adversarial Attacks on Deep-Learning Based Radio Signal Classification
• A Hybrid Training-time and Run-time Defense Against Adversarial Attacks in Modulation Classification
• Mitigation of Adversarial Examples in RF Deep Classifiers Utilizing AutoEncoder Pre-training
• Channel-Aware Adversarial Attacks Against Deep Learning-Based Wireless Signal Classifiers