# Unpriortized Autoencoder For Image Generation

## TL;DR

This paper introduces a novel autoencoder-based image generation method that explicitly estimates the latent distribution using a latent density estimator, resulting in improved image quality over previous models.

## Contribution

It proposes a new approach to image generation with autoencoders by directly estimating the latent distribution, bypassing manual prior assumptions.

## Key findings

- Generated images have higher visual quality.
- The model outperforms previous autoencoder-based generative models.
- Explicit latent distribution estimation improves generation results.

## Abstract

In this paper, we treat the image generation task using an autoencoder, a representative latent model. Unlike many studies regularizing the latent variable's distribution by assuming a manually specified prior, we approach the image generation task using an autoencoder by directly estimating the latent distribution. To this end, we introduce 'latent density estimator' which captures latent distribution explicitly and propose its structure. Through experiments, we show that our generative model generates images with the improved visual quality compared to previous autoencoder-based generative models.