Pixyz: a library for developing deep generative models

Abstract

With the recent rapid progress in the study of deep generative models (DGMs), there is a need for a framework that can implement them in a simple and generic way. In this research, we focus on two features of the latest DGMs: (1) deep neural networks are encapsulated by probability distributions and (2) models are designed and learned based on an objective function. Taking these features into account, we propose a new DGM library called Pixyz. We experimentally show that our library is faster than existing probabilistic modeling languages in learning simple DGMs and we show that our library can be used to implement complex DGMs in a simple and concise manner, which is difficult to do with existing libraries.