ADS-detector: An attention-based dual stream adversarial example detection method

Sensen Guo, Xiaoyu Li, Peican Zhu, Zhiying Mu

Abstract: Highlights•Adversarial example detection based on prediction confidence and attention mechanism.•A dual-stream framework is designed to detect both slight and large perturbations.•Spatial filters and attention mechanisms are utilized to enhance detection performance.•Extensive experiments have demonstrated the effectiveness of our method.