Reproduce “Explainable Artificial Intelligence (XAI) on TimeSeries Data: A Survey” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Thomas Rojat, Raphaël Puget, David Filliat, Javier Del Ser, Rodolphe Gelin, Natalia Díaz-Rodríguez
This survey provides a comprehensive overview of Explainable Artificial Intelligence (XAI) methods applied to time series data, categorizing existing techniques and highlighting challenges in evaluating explanations for temporal patterns. It also discusses how XAI can enhance the confidence, stability, and robustness of AI models in critical applications like autonomous driving and medicine.
Key insights
- Many XAI methods applied to time series are adaptations from computer vision and natural language processing, highlighting a need for more time-series-specific XAI techniques.
- XAI's role extends beyond basic interpretability, contributing to the confidence, stability, and robustness of AI systems, especially in safety-critical domains.
- There is a gap in developing user- and decision-maker-centric explanations, with most current methods serving developers, alongside a lack of standardized evaluation metrics for time series explanations.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.