DisDiff: Unsupervised Disentanglement of Diffusion Probabilistic Models

01/31/2023
by   Tao Yang, et al.
0

In this paper, targeting to understand the underlying explainable factors behind observations and modeling the conditional generation process on these factors, we propose a new task, disentanglement of diffusion probabilistic models (DPMs), to take advantage of the remarkable modeling ability of DPMs. To tackle this task, we further devise an unsupervised approach named DisDiff. For the first time, we achieve disentangled representation learning in the framework of diffusion probabilistic models. Given a pre-trained DPM, DisDiff can automatically discover the inherent factors behind the image data and disentangle the gradient fields of DPM into sub-gradient fields, each conditioned on the representation of each discovered factor. We propose a novel Disentangling Loss for DisDiff to facilitate the disentanglement of the representation and sub-gradients. The extensive experiments on synthetic and real-world datasets demonstrate the effectiveness of DisDiff.

READ FULL TEXT

page 3

page 7

page 8

research
06/09/2020

Interpretable Deep Graph Generation with Node-Edge Co-Disentanglement

Disentangled representation learning has recently attracted a significan...
research
12/26/2022

Unsupervised Representation Learning from Pre-trained Diffusion Probabilistic Models

Diffusion Probabilistic Models (DPMs) have shown a powerful capacity of ...
research
09/18/2023

Gradpaint: Gradient-Guided Inpainting with Diffusion Models

Denoising Diffusion Probabilistic Models (DDPMs) have recently achieved ...
research
03/15/2022

Unpaired Deep Image Dehazing Using Contrastive Disentanglement Learning

We present an effective unpaired learning based image dehazing network f...
research
11/04/2022

Modeling Temporal Data as Continuous Functions with Process Diffusion

Temporal data like time series are often observed at irregular intervals...
research
08/23/2023

High-quality Image Dehazing with Diffusion Model

Image dehazing is quite challenging in dense-haze scenarios, where quite...
research
08/26/2020

Orientation-Disentangled Unsupervised Representation Learning for Computational Pathology

Unsupervised learning enables modeling complex images without the need f...

Please sign up or login with your details

Forgot password? Click here to reset