Transferable Universal Adversarial Perturbations Using Generative Models.
Atiye Sadat HashemiAndreas BärSaeed MozaffariTim FingscheidtPublished in: CoRR (2020)
Keyphrases
- generative model
- probabilistic model
- mixture model
- discriminative models
- discriminative learning
- em algorithm
- conditional random fields
- prior knowledge
- expectation maximization
- semi supervised
- hidden variables
- maximum entropy principle
- topic models
- object categories
- representational power
- hierarchical hidden markov models
- object parts
- information extraction
- generative and discriminative models