Using Skew to Assess the Quality of GAN-generated Image Features

TL;DR

This paper introduces the Skew Inception Distance (SID), a new metric that extends FID by incorporating third-moments to better evaluate GAN-generated image features, aligning more closely with human perception.

Contribution

The paper proposes SID, a novel evaluation metric for GANs that accounts for skewness in feature distributions, extending the FID measure and demonstrating its effectiveness.

Findings

SID correlates more closely with human perception in some cases.

SID extends FID by incorporating third-moments.

Principal component analysis speeds up SID computation.

Abstract

The rapid advancement of Generative Adversarial Networks (GANs) necessitates the need to robustly evaluate these models. Among the established evaluation criteria, the Fr\'{e}chetInception Distance (FID) has been widely adopted due to its conceptual simplicity, fast computation time, and strong correlation with human perception. However, FID has inherent limitations, mainly stemming from its assumption that feature embeddings follow a Gaussian distribution, and therefore can be defined by their first two moments. As this does not hold in practice, in this paper we explore the importance of third-moments in image feature data and use this information to define a new measure, which we call the Skew Inception Distance (SID). We prove that SID is a pseudometric on probability distributions, show how it extends FID, and present a practical method for its computation. Our numerical experiments support that SID either tracks with FID or, in some cases, aligns more closely with human perception when evaluating image features of ImageNet data. Our work also shows that principal component analysis can be used to speed up the computation time of both FID and SID. Although we focus on using SID on image features for GAN evaluation, SID is applicable much more generally, including for the evaluation of other generative models.