EXPLAINABLE AI MODELS FOR TIME-SERIES SIGNAL ANALYSIS

Abstract

Time-series signal analysis plays a vital role in applications such as biomedical monitoring, fault diagnosis, speech processing, financial forecasting, and industrial automation. While deep learning models have achieved remarkable performance in analyzing complex temporal patterns, their black-box nature limits trust, transparency, and adoption in safetycritical systems. Explainable Artificial Intelligence (XAI) has emerged as a promising solution to enhance interpretability without sacrificing predictive accuracy. This paper presents a comprehensive study on explainable AI models for time-series signal analysis, focusing on interpretable deep learning architectures and post-hoc explanation techniques. The proposed framework integrates explainability mechanisms with machine learning-based signal processing to provide transparent decision-making. Simulation-based analysis demonstrates that XAI-enabled models improve trust, reliability, and usability while maintaining competitive performance, making them suitable for realworld time-series applications.