Generative Model Adversarial Training for Deep Compressed Sensing

TL;DR

This paper introduces a robust deep generative model for compressed sensing that is resistant to adversarial perturbations, supported by theoretical analysis and validated through real-world experiments.

Contribution

It proposes a novel low-to-high dimensional generator design for deep compressed sensing with robustness to adversarial attacks in the latent space.

Findings

Enhanced robustness of the generator against adversarial perturbations

Theoretical proof of robustness properties

Successful application on real-world datasets

Abstract

Deep compressed sensing assumes the data has sparse representation in a latent space, i.e., it is intrinsically of low-dimension. The original data is assumed to be mapped from a low-dimensional space through a low-to-high-dimensional generator. In this work, we propound how to design such a low-to-high dimensional deep learning-based generator suiting for compressed sensing, while satisfying robustness to universal adversarial perturbations in the latent domain. We also justify why the noise is considered in the latent space. The work is also buttressed with theoretical analysis on the robustness of the trained generator to adversarial perturbations. Experiments on real-world datasets are provided to substantiate the efficacy of the proposed \emph{generative model adversarial training for deep compressed sensing.}