-
3. From Testing to Defenses: When Realistic Attacks Help
Finding the attacks that teach models to defend themselves.
-
2b. Constrained Adversarial DGMs (C-AdvDGMs): Realistic Attacks with Generative Models
Extending C-DGMs to Safely Test Model Vulnerabilities
-
2a. From Deep Generative Models (DGMs) to Constrained Deep Generative Models (C-DGMs)
How to Make Synthetic Tabular Data Realistic by Design
-
2. Realistic Adversarial Attacks for Tabular Data Using Generative Models
-
1. The Limits of Adversarial Text Attacks
When AI Gets Fooled but Humans Don’t.