Trust but Verify: An Information-Theoretic Explanation for the Adversarial Fragility of Machine Learning Systems, and a General Defense against Adversarial Attacks.
Jirong YiHui XieLeixin ZhouXiaodong WuWeiyu XuRaghuraman MudumbaiPublished in: CoRR (2019)
Keyphrases
- information theoretic
- machine learning systems
- mutual information
- information theory
- bregman divergences
- jensen shannon divergence
- theoretic framework
- information theoretic measures
- entropy measure
- machine learning
- learning systems
- kullback leibler divergence
- text mining
- machine learning algorithms
- relative entropy
- support vector
- learning algorithm