WAKE FOREST UNIVERSITY
DEPARTMENT OF MATHEMATICS & STATISTICS
Dr. Erhardt, Dr. Erway, Dr. Hepler, Dr. Jiang
November 15, 2018
Manchester Hall, Room 020
Title: Flooded with Data.
Abstract: Climate change is leading to rising sea levels, an increase in overall precipitation, and stronger storm surges. This all means increased risk of flooding. How can we better utilize existing databases on floods, insurance claims, and climate models to measure this increased risk?
General research interests: Measuring and managing climate change risks. I am interested in projects which utilize large environmental and climate databases to provide accurate estimates of the changing risks of floods, heatwaves, droughts, and other environmental drivers of losses. I am also interested in translating these analyses into actionable business and policy decisions, to help communities better adapt to a world with rising climate risk. This work is often very computational, and affords students extensive opportunities with statistical programming.
Title: Optimization in Deep Learning
Abstract: This talk will give a 3-minute overview of deep learning. We will briefly discuss why optimization is so important to deep learning. An overview of optimization methods will be presented, as well as a hands-on example of deep learning in action. Finally, there will be some discussion on how students can get involved.
Title: Spatio-temporal modeling in ecology and epidemiology
Abstract: My research revolves around development and assessment of spatio-temporal models applied to problems in ecology and epidemiology. In this talk, I will outline two specific applications I work with: 1) modeling presence/absence camera trap data of species in Serengeti National Park and 2) developing statistical models to quantify the opioid epidemic.
Title: Properties of Anisotropic Sobolev Norms
Abstract: In the space of differentiable functions over a closed domain such as a rectangle, various norms are introduced for different purposes. For example, the Sobolev norm, which is a sum of the L^p norms of a function and its derivatives, is used to solve partial differential equations. Recently,
new norms are introduced where the L^p norm of the derivatives is not taken over the entire domain, but just along one direction on a subset. These norms are called anisotropic Sobolev norms. These norms are introduced to study properties of chaotic dynamical systems. My current research interest is focused on obtaining a good understanding of this type of norms and its application to infinite dimensional smooth dynamical systems.