1 paper · 1 filter
Abdullah Ali, Birhanu Eshete
An adversary who aims to steal a black-box model repeatedly queries the model via a prediction API to learn a function that approximates its decision boundary. Adversarial approxim…