Parameterizing the Reverse Process with Neural Networks
Was this section helpful?
Denoising Diffusion Probabilistic Models, Jonathan Ho, Ajay Jain, Pieter Abbeel, 2020Advances in Neural Information Processing Systems (NeurIPS)DOI: 10.48550/arXiv.2006.11239 - This paper introduces the specific parameterization strategy, the noise prediction network, and the fixed variance choices that are central to the section.
Deep Unsupervised Learning using Nonequilibrium Thermodynamics, Jascha Sohl-Dickstein, Eric A. Weiss, Niru Maheswaranathan, Surya Ganguli, 2015International Conference on Machine Learning (ICML)DOI: 10.48550/arXiv.1503.03585 - This foundational paper introduces Diffusion Probabilistic Models, defining the forward and reverse processes and the general framework of approximating the reverse transition.