Primary Menu

Education, Events, Publication

Funding & Recognition

Applying Statistical Distance Metrics for Dimensionality Reduction

Year: 2023


Presenter Name: Anna Bell

Description
At the University of Utah's Department of Chemistry, the Sigman Lab is searching for "hotspots," from experiments involving organic compounds and hundreds machine-learning generated descriptors of these compounds. A "hotspot" may be described as densely concentrated target reactions resulting from experimental asymmetric catalysis. In this project, we attempt to identify meaningful 2-dimensional feature spaces containing these hotspots. Using density measures, we score the "clustering" of highly reactive outcomes and their variance from all other outcomes within each pairwise combination of descriptors. The meaningfulness of the score is then determined using a permutation method which calculates whether a hotspot's distribution differs significantly from the overall distribution of outcomes in its respective 2-dimensional feature space.
University / Institution: University of Utah
Type: Poster
Format: In Person
Presentation #B3
SESSION B (10:45AM-12:15PM)
Area of Research: Engineering
Faculty Mentor: Jeff Phillips