FLEXtime: Filterbank learning to explain time series

TL;DR

FLEXtime introduces a novel approach to time series explainability by decomposing signals into frequency bands with bandpass filters and learning the optimal combination to explain model predictions, outperforming existing methods.

Contribution

The paper presents FLEXtime, a new method that leverages signal decomposition for more interpretable and effective time series explanations, filling a gap in current methodologies.

Findings

FLEXtime outperforms state-of-the-art explainability methods on multiple datasets.

The method effectively interprets complex signals like EEG and audio.

FLEXtime provides more meaningful explanations by focusing on frequency bands.

Abstract

State-of-the-art methods for explaining predictions from time series involve learning an instance-wise saliency mask for each time step; however, many types of time series are difficult to interpret in the time domain, due to the inherently complex nature of the data. Instead, we propose to view time series explainability as saliency maps over interpretable parts, leaning on established signal processing methodology on signal decomposition. Specifically, we propose a new method called FLEXtime that uses a bank of bandpass filters to split the time series into frequency bands. Then, we learn the combination of these bands that optimally explains the model's prediction. Our extensive evaluation shows that, on average, FLEXtime outperforms state-of-the-art explainability methods across a range of datasets. FLEXtime fills an important gap in the current time series explainability methodology and is a valuable tool for a wide range of time series such as EEG and audio. Code is available at https://github.com/theabrusch/FLEXtime.