Collected sources and patterns will appear here. Add from search, explore, or the patterns library.
(Tensor, Model, GradientBound) -> Tensor
Generate adversarial samples by iteratively computing the gradient of the loss function and perturbing the original inputs within an epsilon-bound norm constraint.
Problem it solves
Neural networks are vulnerable to small, human-imperceptible input modifications that shift classifications.
Consumes
Emits
The real projects this mechanism was found in. Attribution is the point — this is how the best teams actually do it.