Department of Mathematics

CMSE 890-003: Computational Inverse Problems

Instructor: Mark Iwen
Time and Place: Tue Thu : 8:30 am -- 9:50 am in Wells Hall A303
E-mail: markiwen@math.msu.edu
Office: D220 WH
Office Hours: Tue Thu : 10 am -- 11 am, and by appointment

In this course we will discuss the fundamentals of inverse problems encountered in science and engineering. We will explore traditional approaches for solving these problems, including linear regression, Fourier techniques, and the Bayesian Method. Additionally, we will also cover contemporary Machine Learning (ML) techniques, such as neural networks and generative priors, used in various reconstruction algorithms. Emphasis will be placed on understanding the theory and mathematics behind Standard and ML methods for inverse problems. Our primary focus will be on imaging applications, specifically natural image processing.

Course website for CMSE890-003:

http://math.msu.edu/~markiwen/Teaching/CMSE890/CMSE890_S25.html

The course website has the syllabus.

Textbook:

This course has no textbook. Instead we will be covering material from several papers and book chapters during the semester. These resources are linked to on the course's D2L page.

Paper Presentations:

Students are expected to develop and then present lectures on one or two research papers (approved in advance by the instructor). It is anticipated that each student's lecture will summarize a paper's content and answer questions over the course of roughly 40 -- 60 minutes. Alternatively, two students may work together to present a longer paper (or two related shorter papers) over the course of an entire 80 minute lecture. A list of suggested papers will be posted on the course's D2L page.

Class Participation:

Please ask questions, answer questions, make constructive comments, and generally "show your face" by attending class.

Grading:

Your final grade will be assigned based on your paper presentation, as well as on your participation (i.e., lecture attendance).