Assistant Prof. Changyue Song gave a lecture titled “Data Science for Status Inference and Prediction in Smart Systems” at AMSS on April, 5, 2021.
This talk focuses on data science for status inference and prediction in smart systems with applications in crowdsourcing and predictive maintenance. Crowdsourcing has been a prompt and cost-effective way of solving a large number of intelligence-intensive tasks by assigning the tasks to anonymous online workers. Since workers are unreliable, usually one task is assigned to multiple workers and obtain multiple responses or labels. To infer the correct label, a number of models have been developed in the literature that aggregating the labels given by the workers. However, most of the existing studies assume workers to be independent, and thus are sensitive to worker collusion. In this talk, a novel statistical model will be introduced to simultaneously detect collusion, learn workers’ expertise, and infer the correct labels, by treating workers in a pairwise manner. Theoretical properties and numerical experiments that verify the accuracy in parameter estimation and collusion detection will be discussed. Another research topic that will be introduced in this talk is condition monitoring and predictive maintenance where multiple sensors are used to simultaneously monitor the degradation process of a single unit. A novel health index method and several extensions related to the topic will be presented to infer the underlying degradation status and predict the failure time of the unit.
Changyue Song is an assistant professor in the School of Systems and Enterprises at Stevens Institute of Technology. His research interests include data analytics for system improvement, process modeling and prognosis, and statistical learning. He received the Ph.D. degree in industrial engineering and the M.S. degree in statistics from the University of Wisconsin-Madison in 2020. He is the recipient of the 2019 INFORMS QSR Section Best Referred Paper Award, Wisconsin Distinguished Graduate Fellowship, and Mary G. and Joseph Natrella Scholarship.
