Universal Fourier Attack for Time Series

• Published in: IEEE open journal of signal processing 2024-01, Vol.5, p.858-866
• Main Authors: Coda, Elizabeth, Clymer, Brad, DeSmet, Chance, Watkins, Yijing, Girard, Michael
• Format: Article
• Language: English
• Subjects: Adversarial machine learning; artificial intelligence; Audio data; automatic speech recognition; Data models; deep learning; Digital imaging; Fourier series; Frequency spectrum; Image manipulation; Predictive models; Recording; Time series; Time series analysis; Time-frequency analysis; Training; Wireless networks
• ISSN: 2644-1322; 2644-1322
• DOI: 10.1109/OJSP.2024.3402154

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 defenses. We demonstrate the effectiveness of the attack on two different classification tasks through both digital and real world experiments, and show that the attack is robust against common transform-and-compare defense pipelines.