-演講時間: 114年12月3日（三） 14:00~16:00

-演講地點: E6-A207教室

-演講者: 陳家正 助理教授 (元智大學工業工程與管理學系所)

-演講主題: Explainable AI and It's Applications in Semiconductor Manufacturing Industry

-演講摘要：With the advancement of smart manufacturing in the semiconductor industry, big data analytics and machine learning technologies are continuously evolving, especially in the analysis of high-dimensional and highly collinear process parameters. Such data structures often pose challenges for parameter estimation in multivariate analysis, adversely affecting both model interpretability and predictive accuracy. In semiconductor manufacturing, certain collinear variables, despite their interdependence, play critical roles in explaining key process outcomes and therefore cannot be easily removed or ignored.

This study aims to explore and develop a novel AI modeling approach that integrates domain knowledge to improve the accuracy and interpretability of parameter estimation under collinearity. The proposed method is designed to support smarter and more transparent decision-making in manufacturing processes. Empirical results demonstrate that the approach significantly enhances the interpretability of regression coefficients and provides more accurate and physically meaningful estimates for collinear variables.