ECE grad student finalist for best paper at ACC 2022
A graduate student in Purdue University’s Elmore Family School of Electrical and Computer Engineering was a finalist for the best student paper award at the 2022 American Control Conference. Lei Xin is a student of Shreyas Sundaram, Marie Gordon Associate Professor of Electrical and Computer Engineering.
The title of Xin’s paper is “Identifying the Dynamics of a System by Leveraging Data from Similar Systems.” In order to control a system (i.e., make it behave in a desired manner), it is often useful to first formulate a mathematical model of that system (i.e., understand how it responds when inputs are applied to it). One way to learn the model is by interacting with the system and gathering data from it. However, gathering data from a single system is often difficult, time-consuming, or expensive. Instead, one may have lots of data from other similar systems. The key question is then how to utilize all of this additional data from other systems to help learn the model of the original system.
This paper answers this question by providing an approach to blend data from multiple different systems to most effectively learn the model for each of those systems. This work was motivated by a precision animal agriculture project, where the goal is to model dairy cattle (in order to better choose their feed and improve their health, milk production, and greenhouse gas impacts). In that project, it is very challenging to obtain a lot of data from any individual animal; our work shows how to leverage data from other animals in the herd, as well as reasonable simulators of dairy cow metabolic processes, to learn the model of each individual animal.
The American Control Conference (ACC) brings together an international community of researchers and practitioners to discuss the latest findings in automatic control. ACC 2022 was held June 8-10 in Atlanta, GA.