|Date: Friday, February 23, 2018
Location: 3088 East Hall (3:00 PM to 4:00 PM)
Title: Academic Careers @ Liberal Arts Colleges: Research Presentations
Abstract: Research Presentations:
Maria Birgen, Professor of Mathematics, Wartburg College
Abstract: At a small liberal arts college, there is not a lot of time to dedicate to Research, or even research. However, that does not mean that you are doomed to intellectual stagnation. I balance my research time between teaching innovation and supervising undergraduate mathematics projects that the students develop. This means I read about research on how students learn, learn to program in new languages, and learn just enough to be dangerous in a variety of mathematical fields. In particular, I have written the equivalent of a textbook for my Liberal Arts class, the Mathematics of Democracy.
Marie Snipes, Associate Professor of Mathematics and Statistics, Kenyon College
Title: Fun with Fractals and Metric Space Embeddings
Abstract: Metric spaces are useful for modeling sets for which there is a natural notion of distance; examples include geographical data, genetic data, and signals (like sound waves or images). We often seek to classify metric spaces by how Euclidean they look: specifically, we want to know if some metric space can be embedded in Euclidean space. In this talk we will use examples to introduce the general embedding problem, building up to the fractal metric space constructed by Laakso. We will then discuss a theorem by Assouad guaranteeing bi-Lipschitz embeddings of snowflake metric spaces into Euclidean space and related work by Naor-Nieman and David-Snipes.
Darin R. Stephenson, Professor of Mathematics, Hope College
Title: Using Convolutional Neural Networks to Identify Bird Species from Birdsong Samples
Darin Stephenson (joint work with Russell Houpt, Sarah Seckler, Paul Pearson, and Mark Pearson)
Abstract: Extending 2016 work by Hope Colege students Alli VanderStoep and Taylor Rink, our program aims to develop a computer algorithm for determining the species of certain birds given an audio recording of birdsong. Our current data set consists of over 24,000 audio recordings coming from nearly 1,000 different species. I will describe our pre-processing steps which involve the use of real and complex wavelets to denoise birgsong signal and extract time-frequency scalograms. Once the scalograms are developed, we use image recognition techniques to build and train a convolutional neural network to match scalograms with bird species.
Speaker: Mariah Birgen, Marie Snipes, Darin Stephenson
Event Organizer: Karen Smith email@example.com