Broadly speaking, my research interests are in the intersection of computer science, mathematics & statistics, and domain applications. More specifically, my primary research interests are in data science, machine learning, scientific computing, and high-performance computing. In particular, I am interested in the design and application of computationally efficient methods for modeling and solving real world problems in domain areas.

Compression Method

My dissertation, advised by Michael Heath, focused on developing a new numerical method for solving discretized linear system. This method capitalized on the idea of compression, recognizing that the most convenient basis for the discretization may not yield a compressed representation of the soluton. The compression method instead incrementally looks for a solution in a reduced space, finding coefficients to discrete basis vectors. I also developed a multi-faceted stopping criterion to detect when the basis was large enough to generate a solution of satisfactory accuracy.

Ongoing work is examining the applicability of this method to problems whose true solutions have more complex behavior and might warrant a customized basis.

Recent Student Research Projects

Much of my research is very student focused. I work closely one-on-one with a student to find a mutually interesting topic for a project. Below are some of my recent/ongoing student research projects.

Student Researcher: Seth Ockerman | 2020 - Present

This project focuses on using social media image data to predict COVID-19 case counts. Seth created a large social media image dataset for detecting the presence of face masks in images, trained a CNN to detect the presence of face masks, and has been performing time series analysis to develop predictive models of COVID case counts. This project has been funded at various stages by the GVSU Student Summer Scholars (S3) grandt and the MI-STEM Forward grant.

Fast Gaussian Process Emulation of Mars Global Climate Model

Co-mentored with Dr. Nathaniel Bowman (GVSU)

Student Researcher: Marc Tunnell | 2022 - Present

This project focused on emulating the Mars Global Climate Model (MGCM) using gaussian processes. Marc created a surrogate model that emulates the MGCM after training on only a few parameters. This allows us to replicate sensitivity studies that have traditionally been extremely computationally taxing in far less time, even potentially opening up the opportunity for more fine-grained sensitivity studies. Marc’s work on this project was funded at various stages by the GVSU Office of Undergraduate Research and Scholarship Kindschi Fellowship and the MI-STEM Forward grant.

An Augmented Image Captioning Model: Incorporating Hierarchical Image Information

Co-mentored with Dr. Greg Wolffe (GVSU)

Student Researcher: Nathan Funckes | 2020 - 2021

This project focused on improving the accuracy of automated image captioning. Specifically, Nathan investigated whether incorporating category labels (i.e. high-level object superclasses) produced a statistically significant increase in the quality (both quantitatively and qualitatively) of the generated captions. This project was funded by the GVSU Modified Student Summer Scholars (MS3) grant.

Nathan is currently enrolled in the PhD program at Oregon State University.