Low-Dimensional Learning for System Monitoring and Control Using High-Dimensional Data Streams

Industry 4.0, along with advancements in sensing and communication, has enabled the large-scale collection of streaming data, creating unique opportunities for system modeling and monitoring. However, the complex nature of these datasets presents significant analytical challenges. Common characteristics include high variety, high dimensionality, high velocity, and intricate spatial and temporal structures. In this talk, I will present our research on developing efficient methods for system monitoring and control using high-dimensional data streams. The proposed frameworks leverage low-dimensional representations of high-dimensional data and can accommodate various data types, including profiles, images, videos, point clouds, and manifolds. These methods have been validated across multiple application domains, such as additive manufacturing, automotive, forging and rolling, and environmental monitoring.

Bio: Kamran Paynabar is the Fouts Family Chair and Professor in the H. Milton Stewart School of Industrial and Systems Engineering at Georgia Tech. His research focuses on methodological and applied aspects of statistical machine learning for engineering applications, supported by NSF, NIH, DOE, and industry leaders such as Samsung, Ford, and Boeing. He has received best paper awards from INFORMS, IISE, ASA, and POMS, along with multiple teaching honors. He is Editor-Elect of Technometrics and former Department Editor for IISE Transactions. A Fellow of ASQ and elected ISI member, he also co-founded ProcessMiner, an AI-driven manufacturing analytics company.