Great lecture and paper. Now I've got to find the new developments, especially in the parallels with Simoncelli's work on BF-CNN and denoisers and diffusion since then. An excellent paper coming from the other side is "Generalization in diffusion models arises from geometry-adaptive harmonic representation"
Great lecture and paper. Now I've got to find the new developments, especially in the parallels with Simoncelli's work on BF-CNN and denoisers and diffusion since then. An excellent paper coming from the other side is "Generalization in diffusion models arises from geometry-adaptive harmonic representation"
Very nice work.
This sound groundbreaking, or am I looking into this too much?