In the evolving landscape of data science, the critical challenges are not only to train accurate models but also to ensure effective decision-making, explainability, and the trustworthiness of AI systems for real-world applications. This talk will focus on two topics.

First, I will showcase applications in manufacturing data science, where prediction integrated with decision-making approaches—such as reinforcement learning—enables energy forecasting and management and preventive/predictive maintenance in intelligent manufacturing systems.

Second, I will introduce advances in explainable AI (XAI), emphasizing methodological advancements—such as enhanced local interpretability techniques—that improve the fidelity, stability, and reliability of explanations.

Through these perspectives, I will illustrate how data science can evolve beyond prediction toward trustworthy AI systems, bridging the gap between scientific advancement and practical application.