Zahra Kadkhodaie — Blind Denoising Diffusion Models (BDDM) and Adaptive Sampling Algorithm

Denoising diffusion models (DDMs) are state-of-the-art for numerous tasks pertaining to image generation and inverse problems. Yet many aspects of the training and sampling pipeline remain poorly understood. The necessity of noise conditioning has been particularly elusive, forcing practitioners to incorporate unnatural noise embeddings into neural network architectures and use ad hoc noise schedules during sampling. Motivated to remove these inconsistencies, a complete theory for blind denoising diffusion models (BDDMs) is provided: a variant of DDMs where the noise amplitude is not passed into the neural network during training nor sampling. The correctness of BDDMs as a sampling algorithm is justified under the sole assumption of low intrinsic dimensionality of the underlying data distribution relative to the ambient dimension.