Assistant Professor Hao Yan gave a lecture titled ‘High-dimensional data modeling and Monitoring via Knowledge-induced Deep Learning ‘on Jan. 9, 2021, at AMSS.
The wide accessibility of imaging, profile sensors, or text information in modern industrial systems has led to a growing interest in monitoring and modeling high-dimensional (HD) data. The deep learning method has been quite popular recently, given its excellent performance and accuracy by utilizing a large amount of HD data. However, deep learning methods typically require tons of samples and lack interoperation power. Therefore, they are not widely used in realistic industry problems. We will present three relevant works in deep learning for anomaly detection and system modeling addressing these challenges. First, they will present our work of utilizing variational autoencoders for nonlinear profile monitoring with proposed monitoring statistics. They will also discuss how these monitoring statistics naturally extend those for PCA methods. Second, they will present our work in sequential failure event prediction in the aviation system, where thousands of potential failure events may occur. They showed that by exploring the hierarchical tree structures of the failure events, the standard deep learning methods' accuracy could be improved with better interpretability. Third, they will discuss our work on incorporating the anatomical constraint into the lesion detection in dental CBCT imaging. They showed that by incorporating the domain knowledge, the lesion detection accuracy could be significantly improved.
Hao Yan received the B.S. degree in physics from Peking University, Beijing, China, in 2011, and the M.S. degree in statistics, the M.S. degree in computational science and engineering, and the Ph.D. degree in industrial engineering from the Georgia Institute of Technology, Atlanta, GA, USA, in 2015, 2016, and 2017, respectively. He is currently an Assistant Professor with the School of Computing, Informatics, and Decision Systems Engineering (SCIDSE), Arizona State University (ASU), Tempe, AZ, USA. His research interests focus on developing scalable statistical learning algorithms for large-scale high-dimensional data with complex heterogeneous structures to extract useful information for the purpose of system performance assessment, anomaly detection, intelligent sampling, and decision making. Dr. Yan was a recipient of multiple awards, including the Best Paper Award in the IEEE Transactions on Automation Science and Engineering (TASE), IISE Transaction QCRE, and the ASQ Brumbaugh Award. He is a member of INFORMS and IISE.
