Keyizhi Xu

Rethinking the Adversarial Robustness of Multi-Exit Neural Networks in an Attack-Defense Game