AI RESEARCH
Semantic Smoothing via Novel View Synthesis for Robust SAR Image Classification
arXiv CS.CV
•
ArXi:2605.16440v1 Announce Type: new Deep neural networks are vulnerable to adversarial perturbations, limiting deployment in safety-critical applications such as synthetic aperture radar (SAR) automatic target recognition (ATR). Randomized smoothing improves robustness by averaging predictions over noisy inputs, but isotropic noise often fails to preserve the semantic structure of SAR imagery. We propose semantic smoothing, a defense that replaces noised-based perturbations with structured randomized transformations generated by a novel view synthesis model.