ConceptAI & ML
Projected gradient ascent
Periodically projecting the generated image during optimization onto a nicer set of images with properties like spatial smoothness.
Know first
Named as a prerequisite by the lectures that define this. The number is how many say so. A grey one has no page yet.
Lecturers
2Each in their own terms, from the easiest level.
Sara Beery
An iterative method to find an adversarial perturbation by taking gradient steps to maximize the loss and then projecting the result back into a feasible set of allowed perturbations.
Connections
6The lectures that define it or assume it, and the concepts next to it.