We show that AdvCL is able to enhance cross-task robustness transferability without loss of model accuracy and finetuning efficiency.
Physical adversarial attacks apply carefully crafted adversarial perturbations onto real objects to maliciously alter the prediction of object classifiers or detectors.
We further show that our formulation is better at training models that are robust to adversarial attacks.
Recent works have developed several methods of defending neural networks against adversarial attacks with certified guarantees.
Current deep neural networks are vulnerable to adversarial attacks, where adversarial perturbations to the inputs can change or manipulate classification.
When the training data are maliciously tampered, the predictions of the acquired deep neural network (DNN) can be manipulated by an adversary known as the Trojan attack (or poisoning backdoor attack).
Recent works have empirically shown that there exist adversarial examples that can be hidden from neural network interpretability (namely, making network interpretation maps visually similar), or interpretability is itself susceptible to adversarial attacks.
Zeroth-order (ZO) optimization is a subset of gradient-free optimization that emerges in many signal processing and machine learning applications.
Using the framework on images from ImageNet, we demonstrate systems that unify 41% to 46% of predictions in general and unify 67% to 75% of predictions when the systems can explain their conceptual differences.
To the best of our knowledge, this is the first work that models the effect of deformation for designing physical adversarial examples with respect to-rigid objects such as T-shirts.
Recent works have empirically shown that there exist adversarial examples that can be hidden from neural network interpretability, and interpretability is itself susceptible to adversarial attacks.
It is widely known that convolutional neural networks (CNNs) are vulnerable to adversarial examples: images with imperceptible perturbations crafted to fool classifiers.