Full Professor, University of Potsdam
1 paper at NeurIPS 2025
We propose a training-free framework for sampling by exploiting multiscale dynamics and general diffusion paths.