AM-SURE 2023 Project: Optimal Transport for Factor Discovery | Daniel J. King, Kai M. Hung
This repository is a collection of two projects motivated by the usage of optimal transport for factor discovery.
One project is led by Kai M. Hung with a focus on variability reduction via optimal transport. Variability reduction is at the core of factor discovery, since the removal of a factor that maximally reduces variability can be reveal the factor that maximally influences the distribution for some data of interest. Here, we also link to the corresponding report and slides.
Another project is led by Daniel J. King and investigates variants of k-Means clustering using general costs. In the case of a categorical-valued factor, factor discovery becomes a clustering problem. Taking the lense of optimal transport yields variations on k-Means that incorporate new transport costs and initialization schemes. Here, we again link to the corresponding report and slides.
Please navigate into the project subdirectories for an overview of the projects respectively (found in their README.md).