Pros and Cons of GAN Evaluation Measures

TL;DR

This paper reviews and critically discusses over 24 quantitative and 5 qualitative measures for evaluating GANs, highlighting the lack of consensus and proposing desiderata for effective evaluation metrics.

Contribution

It provides a comprehensive review of existing GAN evaluation measures and introduces criteria to assess their effectiveness and suitability.

Findings

Many measures lack consensus and standardization.

Few measures fully meet the proposed evaluation desiderata.

The review highlights the need for improved, standardized evaluation metrics.

Abstract

Generative models, in particular generative adversarial networks (GANs), have received significant attention recently. A number of GAN variants have been proposed and have been utilized in many applications. Despite large strides in terms of theoretical progress, evaluating and comparing GANs remains a daunting task. While several measures have been introduced, as of yet, there is no consensus as to which measure best captures strengths and limitations of models and should be used for fair model comparison. As in other areas of computer vision and machine learning, it is critical to settle on one or few good measures to steer the progress in this field. In this paper, I review and critically discuss more than 24 quantitative and 5 qualitative measures for evaluating generative models with a particular emphasis on GAN-derived models. I also provide a set of 7 desiderata followed by an evaluation of whether a given measure or a family of measures is compatible with them.