Improving Integrated Gradient-based Transferable Adversarial Examples by Refining the Integration Path

Open in new window