Omar Fawzi

Universal Adversarial Perturbations

Given a state-of-the-art deep neural network classifier, we show the existence of a universal (image-agnostic) and very small perturbation vector that causes natural images to be misclassified with high probability. We propose a systematic algorithm for computing universal perturbations, and show that state-of-the-art deep neural networks are highly vulnerable to such perturbations, albeit being quasi-imperceptible to the human eye. We further empirically analyze these universal perturbations and show, in particular, that they generalize very well across neural networks. The surprising existence of universal perturbations reveals important geometric correlations among the high-dimensional decision boundary of classifiers. It further outlines potential security breaches with the existence of single directions in the input space that adversaries can possibly exploit to break a classifier on most natural images.

Quantum Conditional Mutual Information and Approximate Markov Chains

AbstractA state on a tripartite quantum system

Universal recovery map for approximate Markov chains

{A \otimes B \otimes C}

Classical Communication Over a Quantum Interference Channel

A⊗B⊗C forms a Markov chain if it can be reconstructed from its marginal on

On Variational Expressions for Quantum Relative Entropies

{A \otimes B}

From Low-Distortion Norm Embeddings to Explicit Uncertainty Relations and Efficient Information Locking

A⊗B by a quantum operation from B to