Distribution-Interpolation Trade off in Generative Models

Damian Leśniak, Igor Sieradzki, Igor Podolak

Abstract: We investigate the properties of multidimensional probability distributions in the context of latent space prior distributions of implicit generative models. Our work revolves around the phenomena arising while decoding linear interpolations between two random latent vectors -- regions of latent space in close proximity to the origin of the space are oversampled, which restricts the usability of linear interpolations as a tool to analyse the latent space. We show that the distribution mismatch can be eliminated completely by a proper choice of the latent probability distribution or using non-linear interpolations. We prove that there is a trade off between the interpolation being linear, and the latent distribution having even the most basic properties required for stable training, such as finite mean. We use the multidimensional Cauchy distribution as an example of the prior distribution, and also provide a general method of creating non-linear interpolations, that is easily applicable to a large family of commonly used latent distributions.

Keywords: generative models, latent distribution, Cauchy distribution, interpolations

TL;DR: We theoretically prove that linear interpolations are unsuitable for analysis of trained implicit generative models.

Data: [CelebA](https://paperswithcode.com/dataset/celeba)