Search Results for author:
Found 2 papers, 0 papers with code
Adversarial Examples Detection with Enhanced Image Difference Features based on Local Histogram Equalization
However, numerous studies have proven that previous methods create detection or defense against certain attacks, which renders the method ineffective in the face of the latest unknown attack methods.
Adversarial Example Defense via Perturbation Grading Strategy
In addition, the proposed method does not modify any task model, which can be used as a preprocessing module, which significantly reduces the deployment cost in practical applications.
Cannot find the paper you are looking for? You can
Submit a new open access paper.
