Data Dialoue with Chris Trailie

Oct 9

Friday, October 9, 2020 - 12:00pm to 1:00pm

PLEASE REGISTER TO JOIN

Add to calendar »

Presenter

Chris Trailie

My work fuses a passion for hands on coding and math with a desire to explore research topics that are fun, relevant, accessible to undergraduates. As such, I have gravitated towards applications and approaches that are easy to explain and which involve strong visual and interactive components, with the hope of making the traditionally aloof and menacing fields of math and CS more approachable to everyone. For instance, the "shape" of musical audio features can be used to enhance algorithms for audio cover song identification, music structure analysis, and audio alignment (and I get to show heavy metal covers of "Despacito" in the process of explaining this to students: http://www.covers1000.net/demo.html). The notion of a "loop" is also incredibly helpful when analyzing recurrent phenomena, such musical rhythms, but also video data of anomalies in vibrating vocal folds, stereotypical motor motions for autism spectrum disorder patients, and periodic phenomena in p53 gene expression marking cell damage. A few surprising families of shapes, such as torii and Moebius strips, also emerge in these contexts, and the applications help to explain the math in a clearer manner. Finally, I will talk about some recent work on software for doing nonlinear dimension reduction from manifolds with twists, in which undergraduates jointly studying math and CS have found a nice niche application leveraging both of their skillsets. Please register: https://duke.zoom.us/j/96666831540

Contact

Dawn, Ariel
919-684-9312
ariel.dawn@duke.edu