Full Professor, Massachusetts Institute of Technology
1 paper at NeurIPS 2025
We examine the inductive biases of conditional diffusion and devise a rigorous metric for when diffusion models do not behave like denoisers.
