Search

Evaluation measures of Generative Adversarial Networks (GANs) have been an active area of research and, currently, there are several measures to evaluate them. The most used GANs evaluation measure is the Fréchet Inception Distance (FID). Measures such as FID are known as model-agnostic methods, where the generator is used as a black box to sample the generated images. Like other measures of model-agnostic, FID uses a deep supervised model for mapping real and generated samples to a feature space. We proposed an approach here with a deep unsupervised model, the Vector Quantised-Variational Autoencoder (VQ-VAE), for estimating the mean and the covariance matrix of the Fréchet Distance and named it Fréchet AutoEncoder Distance (FAED). Our experimental results highlighted that the feature space of the VQ-VAE describes a clustering domain-specific representation more intuitive and visually plausible than the Inception network used by the benchmark FID. • The latent space of a deep unsupervised model gets a domain-specific representation. • The mean and covariance of Fréchet distance are estimated via a deep unsupervised model. • FAED catches the increased level and magnitude of disturbances on generated images. • The framework projects the data into a lower dimensionality space through the reconstruction error. • The novelty extends to other measures that use a deep supervised model to evaluate GANs. [ABSTRACT FROM AUTHOR]

This paper develops a new integrated ball (pseudo)metric which provides an intermediary between a chosen starting (pseudo)metric d and the L p distance in general function spaces. Selecting d as the Hausdorff or Fréchet distances, we introduce integrated shape-sensitive versions of these supremum-based metrics. The new metrics allow for finer analyses in functional settings, not attainable applying the non-integrated versions directly. Moreover, convergent discrete approximations make computations feasible in practice. [ABSTRACT FROM AUTHOR]