English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Diffusion models are a potent class of generative models capable of producinghigh-quality images. However, they can face challenges related to data bias,favoring specific modes of data, especially when the training data does notaccurately represent the true data distribution and exhibits skewed orimbalanced patterns. For instance, the CelebA dataset contains more femaleimages than male images, leading to biased generation results and impactingdownstream applications. To address this issue, we propose a novel method thatleverages manifold guidance to mitigate data bias in diffusion models. Our keyidea is to estimate the manifold of the training data using an unsupervisedapproach, and then use it to guide the sampling process of diffusion models.This encourages the generated images to be uniformly distributed on the datamanifold without altering the model architecture or necessitating labels orretraining. Theoretical analysis and empirical evidence demonstrate theeffectiveness of our method in improving the quality and unbiasedness of imagegeneration compared to standard diffusion models.

Unbiased Image Synthesis via Manifold-Driven Sampling in Diffusion Models