Fre-CW: Targeted Attack on Time Series Forecasting using Frequency Domain Loss
This addresses a security problem for users of time series forecasting models, but it is incremental as it adapts existing methods to a new domain.
The paper tackles the vulnerability of transformer-based time series forecasting models to adversarial attacks by proposing a targeted attack algorithm using frequency domain loss, achieving excellent performance on major datasets.
Transformer-based models have made significant progress in time series forecasting. However, a key limitation of deep learning models is their susceptibility to adversarial attacks, which has not been studied enough in the context of time series prediction. In contrast to areas such as computer vision, where adversarial robustness has been extensively studied, frequency domain features of time series data play an important role in the prediction task but have not been sufficiently explored in terms of adversarial attacks. This paper proposes a time series prediction attack algorithm based on frequency domain loss. Specifically, we adapt an attack method originally designed for classification tasks to the prediction field and optimize the adversarial samples using both time-domain and frequency-domain losses. To the best of our knowledge, there is no relevant research on using frequency information for time-series adversarial attacks. Our experimental results show that these current time series prediction models are vulnerable to adversarial attacks, and our approach achieves excellent performance on major time series forecasting datasets.