Classes

ME 584 – System Identification

If you are interested in data science or any experimental work, please check this course. ME 584 deals with building models from data or the conditions that experiments must satisfy to extract the parameters of physical systems. The course will also give you some of the rigorous bases of AI/ML methods. You can contact course instructor Dr. Ariyur for questions at kariyur@purdue.edu

ME 584–System Identification

Instructor: Kartik B. Ariyur (https://www.researchgate.net/profile/Kartik_Ariyur)

Course description: The course this time will be suitable to students all around science and engineering and the data sciences. Students will be able to work on a variety of course projects identifying generalized least squares models, or dynamic system models, or various machine learning models. It will provide a fundamental grasp on the underlying theory (least squares, maximum likelihood estimation and connection to experiment design), models for estimation (ARMAX type or difference equation models, neural nets–supervised/unsupervised, physics-based), estimation algorithms (subspace, linear dimensionality reduction, convergence issues) and their implementation (testing for optimality, robustness, and statistical validity). Extensions to inverse problems and pattern recognition will also be covered. The course will possibly have an online section later on for those looking for schedule flexibility.

The course has typically had students from all over engineering, including those from thermal and fluid sciences, food science, and agriculture. See the attached syllabus from last year for the details.