BAGM: A Backdoor Attack for Manipulating Text-to-Image Generative Models.
Jordan ViceNaveed AkhtarRichard HartleyAjmal MianPublished in: CoRR (2023)
Keyphrases
- generative model
- image features
- input image
- multiscale
- image content
- bayesian framework
- probabilistic model
- visual patterns
- image retrieval
- object categories
- image matching
- discriminative learning
- test images
- image representation
- image segmentation
- keypoints
- mixture model
- information retrieval
- em algorithm
- image statistics
- generative and discriminative models
- natural images
- expectation maximization
- image classification
- prior knowledge
- object parts