FREQuency ATTribution: Benchmarking Frequency-based Occlusion for Time Series Data

2025-06-23

AI Summary

To address the limited interpretability of deep time-series models, this paper introduces FreqATT—the first frequency-domain attribution framework tailored for time-series data. Unlike mainstream time-domain explanation methods, FreqATT leverages Fourier transformation to project inputs into the frequency domain, then integrates frequency importance scoring with targeted spectral masking to precisely identify discriminative frequency components and their corresponding time-domain segments. The framework ensures both robustness—maintaining stable attributions under signal noise, scaling, and temporal deformation—and enhanced interpretability—accurately localizing predictive time-series patterns. Extensive experiments demonstrate that FreqATT consistently outperforms state-of-the-art time-domain attribution methods (e.g., Grad-CAM, Integrated Gradients) across diverse time-series tasks, including classification and anomaly detection. By grounding explanations in the frequency domain, FreqATT establishes a verifiable, traceable, and physically meaningful paradigm for interpreting time-series AI decisions.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Computer Vision: Interpretability, Explainability, and Transparency
• Humans and AI: Explainable AI (XAI) for Human Understanding

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Deep neural networks are among the most successful algorithms in terms of performance and scalability in different domains. However, since these networks are black boxes, their usability is severely restricted due to the lack of interpretability. Existing interpretability methods do not address the analysis of time-series-based networks specifically enough. This paper shows that an analysis in the frequency domain can not only highlight relevant areas in the input signal better than existing methods, but is also more robust to fluctuations in the signal. In this paper, FreqATT is presented, a framework that enables post-hoc networks to interpret time series analysis. To achieve this, the relevant different frequencies are evaluated and the signal is either filtered or the relevant input data is marked.

Problem

• Lack interpretability in deep neural networks for time series
• Existing methods insufficiently analyze time-series-based networks
• Need robust frequency-domain analysis for better signal interpretation

Innovation

• Frequency domain analysis for time series
• FreqATT framework for interpretability
• Filtering and marking relevant frequencies