This academic paper highlights the use of Explainable AI (XAI) and Deep Learning in detecting faults in rolling element bearings. The authors propose a unique, domain-specific feature attribution framework that allows comparison of a model’s logic with expert reasoning. The framework helps validate the trustworthiness and the generalization ability of different deep learning models. The paper underscores the value of signal processing tools in augmenting XAI techniques and provides a template for similar problems. However, the authors also note that the lack of transparency in decision-making models often hinders their real-world deployment.
Publication date: 20 Oct 2023
Project Page: Not specified
Paper: https://arxiv.org/pdf/2310.12967
