David Uminsky - LatMath 2022 - IPAM's Latinx in the Mathematical Sciences Conference

Explore a comprehensive lecture from the LatMath 2022 conference, where David Uminsky shares his academic journey and insights on mathematics and data science. Delve into topics such as the human aspect of mathematics, data science perspectives, and research initiatives at the Data Science Institute. Discover the innovative DS Preceptorship program, a collaboration between UChicago and CCC. Gain knowledge about topological data analysis and epistasis, followed by a refresher on Fourier analysis and its applications in feature extraction. Learn about non-abelian harmonic analysis and orthogonal decomposition, and understand their role in detecting higher-order interactions. Conclude with a summary of current challenges in the field, including the need for good, annotated multi-class datasets and the complexities of expert annotation agreement.

Personal background
Academic Journey
Math is a human endeavor
Math & Data
Our Perspective on Data Science
DSI Research Initiatives
DSI - Education
One example: DS Preceptorship. UChicago and CCC
Topological Data Analysis
What is Epistasis?
Types of Epistasis?
Algebra? Fourier Analysis - Brief refresher
Fourier analysis as a feature extractor
Fourier analysis - transform
Fourier transform summary
Processing data
Non-abelian harmonic analysis
Orthogonal Decomposition
Our Results: detecting higher order interaction
Summary and directions
Challenge #1 - Good, annotated, multi-class data set
Expert annotation agreement is challenging