Learning to Generate Heavy-tailed Conditional Distribution via Diffusion Model
Under major revision at Operations Research, 2024
This paper develops a diffusion model-based approach for generating samples from heavy-tailed conditional distributions. The method is particularly useful for applications in finance and risk management where heavy-tailed distributions are prevalent.
Authors: H. Liu†, T. Zhu†, J. He, Z. Zheng (†co-first author)
Presented at 2024 INFORMS Annual Meeting