Full Professor, University of Bristol
1 paper at NeurIPS 2025
We extend the concept of distribution compression to joint and conditional distributions, introducing algorithms to produce compressed sets that approximate the joint or conditional distribution of a target dataset.
