mlsanityData to 7 Oct 2026

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

arXiv 2310.20636 ↗First posted 31 Oct 2023cs.CV cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

The rapid advancement of Generative Adversarial Networks (GANs) necessitates the need to robustly evaluate these models. Among the established evaluation criteria, the Fr'echetInception 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.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 15 of its 33 references that are in the map. First posted on arXiv 31 Oct 2023.

Builds on

15 of 33 references.

References · 15

  1. 2018A Style-Based Generator Architecture for Generative Adversarial Networkscited by 5,986
  2. 2014Deep Learning Face Attributes in the Wildcited by 3,679
  3. 2018A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attackscited by 1,472
  4. 2019StarGAN v2: Diverse Image Synthesis for Multiple Domainscited by 1,095
  5. 2019Improved Precision and Recall Metric for Assessing Generative Modelscited by 925
  6. 2020Reliable Fidelity and Diversity Metrics for Generative Modelscited by 425
  7. 2018Assessing Generative Models via Precision and Recallcited by 401
  8. 2017Are GANs Created Equal? A Large-Scale Studycited by 267
  9. 2021Pros and Cons of GAN Evaluation Measures: New Developmentscited by 121
  10. 2020Leveraging the Feature Distribution in Transfer-based Few-Shot Learningcited by 72
  11. 2019Probabilistic Modeling of Deep Features for Out-of-Distribution and Adversarial Detectioncited by 39
  12. 2019The Shape of Data: Intrinsic Distance for Data Distributionscited by 38
3 more
  1. 2018An Improved Evaluation Framework for Generative Adversarial Networkscited by 26
  2. 2021Evaluating generative networks using Gaussian mixtures of image featurescited by 11
  3. 2021Transfer learning based few-shot classification using optimal transport mapping from preprocessed latent space of backbone neural networkcited by 6

Built on by

Not cited yet.