Mathematics Colloquia and Seminars

Return to Colloquia & Seminar listing

We explore applications of machine learning in three unique disciplines: infection prediction in the swine industry, electrical microgrid control, and retinal genotyping in ophthalmology. We begin by addressing infection prediction in swine populations, a major concern restraining the productivity of swine farms globally. A windowing-based machine learning model is built to predict infection in two real-world swine systems and determine valuable features for infection prediction. Next, we consider the tertiary microgrid control problem under uncertainty and devise reinforcement learning algorithms that outperform rule-based control and model predictive control approaches in high uncertainty scenarios. Finally, the potential of deep learning models to predict genotypes from phenotypic information in retinal fundus images is examined. Vision transformer models are constructed to predict the presence of high-risk allele combinations in the CFH and ARMS2 genes, further utilizing the model to produce insights into the phenotype-genotype relationship in AMD.

Zoom Link: https://ucdavis.zoom.us/j/7231234206?pwd=eFMremRjTFhmZlJSdnBMVVlMbTVpQT09&omn=97095944188