APDMs: Adversarial purification diffusion models for automatic modulation classification

H. Q. Zhang

doi:10.1016/j.sigpro.2025.110249

APDMs: Adversarial purification diffusion models for automatic modulation classification

2025-08-23

0

PRE

AI

DOI：10.1016/j.sigpro.2025.110249

Original

Original request for help

Save

Abstract

En 中文

• Proposing APDMs for adversarial purification in wireless communication signals. • Improving traditional Gaussian noise in DM through a noise optimization strategy. • Suppressing high-frequency components effectively using learnable frequency-domain filters. • Replacing MSE loss with Wasserstein-based PSD and higher-order regularization. • Improving adversarial purification for AMC tasks, enhancing classification accuracy.

Keywords:

APDMs

adversarial purification

frequency-domain filters

Wasserstein loss

AMC tasks

APDMs: Adversarial purification diffusion models for automatic modulation classification

Abstract

Journal

Organization