Universal Fourier Attack for Time Series

TL;DR

This paper introduces a universal, frequency-constrained adversarial attack for time series data that is easy to implement, robust against filtering, and effective across different real-world applications like speech recognition.

Contribution

The paper proposes a novel, universal Fourier-based attack for time series that is time-invariant, frequency-constrained, and effective in real-world scenarios.

Findings

Effective in speech recognition domain

Robust against filtering defenses

Applicable to unintended radiated emission

Abstract

A wide variety of adversarial attacks have been proposed and explored using image and audio data. These attacks are notoriously easy to generate digitally when the attacker can directly manipulate the input to a model, but are much more difficult to implement in the real-world. In this paper we present a universal, time invariant attack for general time series data such that the attack has a frequency spectrum primarily composed of the frequencies present in the original data. The universality of the attack makes it fast and easy to implement as no computation is required to add it to an input, while time invariance is useful for real-world deployment. Additionally, the frequency constraint ensures the attack can withstand filtering. We demonstrate the effectiveness of the attack in two different domains, speech recognition and unintended radiated emission, and show that the attack is robust against common transform-and-compare defense pipelines.