Diffusion models (DMs) are widely used for generating high-quality image...
While it is tempting to believe that data distillation preserves privacy...
We are interested in privatizing an approximate posterior inference algo...
Kernel mean embedding is a useful tool to compare probability measures.
...