Skip to main content
. Author manuscript; available in PMC: 2022 Mar 24.
Published in final edited form as: IEEE Int Workshop Mach Learn Signal Process. 2021 Nov 15;2021:10.1109/mlsp52302.2021.9596214. doi: 10.1109/mlsp52302.2021.9596214

Fig. 3.

Fig. 3.

A) Denoising of MNIST data with alternating (middle row) and integrated (bottom row) diffusion operators. B) MNIST denoised pixel classification accuracy with differing diffusion-based multimodal denoising strategies. As above, data was generated by adding differing amounts of random gaussian noise to MNIST pixels.