Disentangling Disentangled Representations: Towards Improved Latent Units via Diffusion Models

TL;DR

This paper introduces novel techniques to improve unsupervised disentangled representation learning using diffusion models, focusing on interpretability and independence of latent units, achieving state-of-the-art results.

Contribution

It proposes Dynamic Gaussian Anchoring and Skip Dropout to enhance latent disentanglement in diffusion models, addressing interpretability and training challenges.

Findings

State-of-the-art disentanglement on synthetic data

Improved interpretability of latent units

Enhanced downstream task performance

Abstract

Disentangled representation learning (DRL) aims to break down observed data into core intrinsic factors for a profound understanding of the data. In real-world scenarios, manually defining and labeling these factors are non-trivial, making unsupervised methods attractive. Recently, there have been limited explorations of utilizing diffusion models (DMs), which are already mainstream in generative modeling, for unsupervised DRL. They implement their own inductive bias to ensure that each latent unit input to the DM expresses only one distinct factor. In this context, we design Dynamic Gaussian Anchoring to enforce attribute-separated latent units for more interpretable DRL. This unconventional inductive bias explicitly delineates the decision boundaries between attributes while also promoting the independence among latent units. Additionally, we also propose Skip Dropout technique, which easily modifies the denoising U-Net to be more DRL-friendly, addressing its uncooperative nature with the disentangling feature extractor. Our methods, which carefully consider the latent unit semantics and the distinct DM structure, enhance the practicality of DM-based disentangled representations, demonstrating state-of-the-art disentanglement performance on both synthetic and real data, as well as advantages in downstream tasks.