FREQuency ATTribution: Benchmarking Frequency-based Occlusion for Time Series Data
This addresses the lack of interpretability in time-series networks, which restricts usability, though it is incremental as it builds on existing occlusion methods.
The paper tackles the interpretability problem for deep neural networks on time series data by introducing FreqATT, a frequency-based occlusion framework that highlights relevant input areas more effectively and robustly than existing methods.
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.