Boosting Transferability in Vision-Language Attacks via Diversification along the Intersection Region of Adversarial Trajectory.
Sensen GaoXiaojun JiaXuhong RenIvor W. TsangQing GuoPublished in: CoRR (2024)
Keyphrases
- countermeasures
- computer vision
- real time
- language learning
- multi agent
- vision system
- programming language
- input image
- computational linguistics
- weak learners
- image processing
- machine learning
- malicious attacks
- trajectory data
- security threats
- specification language
- security mechanisms
- weak classifiers
- watermarking scheme
- tabu search
- semi supervised
- wireless sensor networks
- natural language