Adaptive Learning of the Latent Space of Wasserstein Generative Adversarial Networks

TL;DR

This paper introduces LWGAN, a framework that adaptively learns the intrinsic data manifold dimension, improving generative quality by combining Wasserstein auto-encoders and GANs, with theoretical guarantees and empirical validation.

Contribution

It proposes a novel LWGAN framework that adaptively learns the data's intrinsic dimension and provides theoretical proofs of its consistency and generalization bounds.

Findings

LWGAN accurately identifies the intrinsic data dimension.

LWGAN generates high-quality synthetic data.

Theoretical guarantees support the method's validity.

Abstract

Generative models based on latent variables, such as generative adversarial networks (GANs) and variational auto-encoders (VAEs), have gained lots of interests due to their impressive performance in many fields. However, many data such as natural images usually do not populate the ambient Euclidean space but instead reside in a lower-dimensional manifold. Thus an inappropriate choice of the latent dimension fails to uncover the structure of the data, possibly resulting in mismatch of latent representations and poor generative qualities. Towards addressing these problems, we propose a novel framework called the latent Wasserstein GAN (LWGAN) that fuses the Wasserstein auto-encoder and the Wasserstein GAN so that the intrinsic dimension of the data manifold can be adaptively learned by a modified informative latent distribution. We prove that there exist an encoder network and a generator network in such a way that the intrinsic dimension of the learned encoding distribution is equal to the dimension of the data manifold. We theoretically establish that our estimated intrinsic dimension is a consistent estimate of the true dimension of the data manifold. Meanwhile, we provide an upper bound on the generalization error of LWGAN, implying that we force the synthetic data distribution to be similar to the real data distribution from a population perspective. Comprehensive empirical experiments verify our framework and show that LWGAN is able to identify the correct intrinsic dimension under several scenarios, and simultaneously generate high-quality synthetic data by sampling from the learned latent distribution.