Analytic-DPM: an Analytic Estimate of the Optimal Reverse Variance in Diffusion Probabilistic ModelsDownload PDF

29 Sept 2021, 00:30 (modified: 03 May 2022, 07:44)ICLR 2022 OralReaders: Everyone
Keywords: diffusion probabilistic models, generative models
Abstract: Diffusion probabilistic models (DPMs) represent a class of powerful generative models. Despite their success, the inference of DPMs is expensive since it generally needs to iterate over thousands of timesteps. A key problem in the inference is to estimate the variance in each timestep of the reverse process. In this work, we present a surprising result that both the optimal reverse variance and the corresponding optimal KL divergence of a DPM have analytic forms w.r.t. its score function. Building upon it, we propose \textit{Analytic-DPM}, a training-free inference framework that estimates the analytic forms of the variance and KL divergence using the Monte Carlo method and a pretrained score-based model. Further, to correct the potential bias caused by the score-based model, we derive both lower and upper bounds of the optimal variance and clip the estimate for a better result. Empirically, our analytic-DPM improves the log-likelihood of various DPMs, produces high-quality samples, and meanwhile enjoys a $20\times$ to $80\times$ speed up.
One-sentence Summary: We propose an analytic framework of estimating the optimal reverse variance in DPMs.
Supplementary Material: zip
21 Replies