Pros and Cons of GAN Evaluation Measures
This work addresses the lack of consensus on evaluation measures for GANs, which is critical for steering progress in generative modeling, but it is incremental as it reviews existing measures rather than proposing new ones.
The paper tackles the problem of evaluating and comparing generative adversarial networks (GANs) by reviewing over 24 quantitative and 5 qualitative measures, and it provides a set of 7 desiderata to assess their compatibility for fair model comparison.
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.