1 paper · 1 filter
Ramchandran Muthukumar, Ambar Pal, Jeremias Sulam +1
State-of-the-art machine learning systems are vulnerable to small perturbations to their input, where ``small'' is defined according to a threat model that assigns a positive threa…