Perturbing the Phase: Analyzing Adversarial Robustness of Complex-Valued Neural Networks

TL;DR

This paper investigates the robustness of complex-valued neural networks against phase-specific adversarial attacks, revealing their vulnerabilities and comparing their robustness to real-valued networks.

Contribution

It introduces Phase Attacks targeting phase information in CVNNs and derives complex-valued adversarial attacks, advancing understanding of their robustness.

Findings

CVNNs can be more robust than RVNNs in some scenarios

Phase Attacks significantly decrease CVNN performance

Both CVNNs and RVNNs are highly susceptible to phase perturbations

Abstract

Complex-valued neural networks (CVNNs) are rising in popularity for all kinds of applications. To safely use CVNNs in practice, analyzing their robustness against outliers is crucial. One well known technique to understand the behavior of deep neural networks is to investigate their behavior under adversarial attacks, which can be seen as worst case minimal perturbations. We design Phase Attacks, a kind of attack specifically targeting the phase information of complex-valued inputs. Additionally, we derive complex-valued versions of commonly used adversarial attacks. We show that in some scenarios CVNNs are more robust than RVNNs and that both are very susceptible to phase changes with the Phase Attacks decreasing the model performance more, than equally strong regular attacks, which can attack both phase and magnitude.