Contextualized Perturbation for Textual Adversarial Attack.
Dianqi LiYizhe ZhangHao PengLiqun ChenChris BrockettMing-Ting SunBill DolanPublished in: NAACL-HLT (2021)
Keyphrases
- keywords
- natural language
- multimedia
- countermeasures
- attack detection
- metadata
- image classification
- intrusion detection
- textual information
- textual data
- detection mechanism
- textual features
- multi agent
- key exchange protocol
- visual content
- secret key
- visual representations
- human generated
- malicious attacks
- attack graph