Projects

Nonparametric Density Estimation using Optimal Transport

I worked with Lorenzo Orecchia to develop a method for estimating (and ultimately sampling from) an unknown probability distribution with compact support in R^n, given a sample from the distribution, while making no strict assumptions about its form. While generative diffusion models solve this problem, our algorithm avoids the use of neural networks and directly utilizes optimal transport and Wasserstein distance. I presented a poster, which you can view here, at UChicago’s 2025 Undergraduate Research Symposium.