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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Deep generative models require large amounts of training data. This oftenposes a problem as the collection of datasets can be expensive and difficult,in particular datasets that are representative of the appropriate underlyingdistribution (e.g. demographic). This introduces biases in datasets which arefurther propagated in the models. We present an approach to construct anunbiased generative adversarial network (GAN) from an existing biased GAN byrebalancing the model distribution. We do so by generating balanced data froman existing imbalanced deep generative model using an evolutionary algorithmand then using this data to train a balanced generative model. Additionally, wepropose a bias mitigation loss function that minimizes the deviation of thelearned class distribution from being equiprobable. We show results for theStyleGAN2 models while training on the Flickr Faces High Quality (FFHQ) datasetfor racial fairness and see that the proposed approach improves on the fairnessmetric by almost 5 times, whilst maintaining image quality. We further validateour approach by applying it to an imbalanced CIFAR10 dataset where we show thatwe can obtain comparable fairness and image quality as when training on abalanced CIFAR10 dataset which is also twice as large. Lastly, we argue thatthe traditionally used image quality metrics such as Frechet inception distance(FID) are unsuitable for scenarios where the class distributions are imbalancedand a balanced reference set is not available.

Fair GANs through model rebalancing for extremely imbalanced class distributions