Efficient Black-box Adversarial Attacks via Bayesian Optimization Guided by a Function Prior

Open in new window