Abstract: Deep neural networks (DNNs)-based SAR target recognition models are susceptible to adversarial examples, which significantly reduce model robustness. Current methods for generating ...