Generative Model without Prior Distribution Matching

TL;DR

This paper introduces a novel generative model that matches the prior to the embedding distribution instead of the latent space, preserving data geometry and improving generation quality.

Contribution

It proposes a new approach where the prior matches the embedding distribution, using a regularized autoencoder and adversarial training to better preserve data structure.

Findings

The method effectively preserves data manifold geometry.

It alleviates the trade-off between reconstruction and generation.

Experimental results support the theoretical advantages.

Abstract

Variational Autoencoder (VAE) and its variations are classic generative models by learning a low-dimensional latent representation to satisfy some prior distribution (e.g., Gaussian distribution). Their advantages over GAN are that they can simultaneously generate high dimensional data and learn latent representations to reconstruct the inputs. However, it has been observed that a trade-off exists between reconstruction and generation since matching prior distribution may destroy the geometric structure of data manifold. To mitigate this problem, we propose to let the prior match the embedding distribution rather than imposing the latent variables to fit the prior. The embedding distribution is trained using a simple regularized autoencoder architecture which preserves the geometric structure to the maximum. Then an adversarial strategy is employed to achieve a latent mapping. We provide both theoretical and experimental support for the effectiveness of our method, which alleviates the contradiction between topological properties' preserving of data manifold and distribution matching in latent space.