Quantitative Evaluation of Generative Models and the GAN Landscape

Generative Adversarial Networks (GANs) are a class of deep generative models which aim to learn a target distribution in an unsupervised fashion. While they were successfully applied to many problems, training a GAN is a notoriously challenging task and requires a significant amount of hyperparameter tuning, neural architecture engineering, and a non-trivial amount of "tricks". The success in many practical applications coupled with the lack of a quantitative measure to quantify the failure modes of GANs resulted in a plethora of losses, regularization and normalization schemes, and neural architectures. In the first part of this talk we will take a sober view of the current state of GANs from a practical perspective and explore the GAN landscape. We will discuss the common pitfalls, provide practical advice for the interested researchers and practitioners. In the second part of the talk we will discuss a novel measure of performance of generative models which is theoretically sound and allows one to distinguish between various failure modes of deep generative models.