Towards a Rigorous Evaluation of XAI Methods on Time Series
This work addresses the need for rigorous evaluation of XAI methods in time series domains, but it is incremental as it applies existing methods to new data with minor adaptations.
The paper tackles the problem of evaluating explainable AI (XAI) methods on time series data, which are typically designed for images and lack rigorous assessment, by introducing new verification techniques and conducting experiments that show SHAP works robustly across models while others like DeepLIFT perform better with specific architectures.
Explainable Artificial Intelligence (XAI) methods are typically deployed to explain and debug black-box machine learning models. However, most proposed XAI methods are black-boxes themselves and designed for images. Thus, they rely on visual interpretability to evaluate and prove explanations. In this work, we apply XAI methods previously used in the image and text-domain on time series. We present a methodology to test and evaluate various XAI methods on time series by introducing new verification techniques to incorporate the temporal dimension. We further conduct preliminary experiments to assess the quality of selected XAI method explanations with various verification methods on a range of datasets and inspecting quality metrics on it. We demonstrate that in our initial experiments, SHAP works robust for all models, but others like DeepLIFT, LRP, and Saliency Maps work better with specific architectures.