2016

Early Methods for Detecting Adversarial Images

Hendrycks, Dan, Gimpel, Kevin

Understand

Many machine learning classifiers are vulnerable to adversarial perturbations.

  • An adversarial perturbation modifies an input to change a classifier's prediction without causing the input to seem substantially different to human perception.
  • We deploy three methods to detect adversarial images.
  • Adversaries trying to bypass our detectors must make the adversarial image less pathological or they will fail trying.

Reading the bibliography…