Poster: Detecting Adversarial Examples Hidden under Watermark Perturbation via Usable Information Theory.
Ziming ZhaoZhaoxuan LiTingting LiZhuoxue SongFan ZhangRui ZhangPublished in: CCS (2023)
Keyphrases
- information theory
- information theoretic
- jensen shannon divergence
- statistical learning
- secret information
- statistical mechanics
- conditional entropy
- kullback leibler divergence
- digital images
- statistical physics
- shannon entropy
- relative entropy
- rate distortion theory
- information hiding
- mdl principle
- mutual information
- digital watermarking
- image processing