Abstract: Although adversarial attacks present a significant threat to intelligent models based on deep learning (DL) for automatic modulation classification (AMC). However, the existing works for ...