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/

Zendy Plus

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

Zendy Tools

Zendy Open

Evaluating image generation models such as generative adversarial networks(GANs) is a challenging problem. A common approach is to compare thedistributions of the set of ground truth images and the set of generated testimages. The Frech\'et Inception distance is one of the most widely used metricsfor evaluation of GANs, which assumes that the features from a trainedInception model for a set of images follow a normal distribution. In thispaper, we argue that this is an over-simplified assumption, which may lead tounreliable evaluation results, and more accurate density estimation can beachieved using a truncated generalized normal distribution. Based on this, wepropose a novel metric for accurate evaluation of GANs, named TREND (TRuncatedgEneralized Normal Density estimation of inception embeddings). We demonstratethat our approach significantly reduces errors of density estimation, whichconsequently eliminates the risk of faulty evaluation results. Furthermore, weshow that the proposed metric significantly improves robustness of evaluationresults against variation of the number of image samples.

TREND: Truncated Generalized Normal Density Estimation of Inception Embeddings for GAN Evaluation