Parallel Multiscale Autoregressive Density Estimation

TL;DR

This paper introduces a parallelized PixelCNN model that significantly speeds up image generation by modeling pixel groups as conditionally independent, enabling practical high-resolution image synthesis with competitive density estimation.

Contribution

The paper presents a novel parallel PixelCNN architecture that reduces sampling complexity from O(N) to O(log N), allowing efficient high-resolution image generation.

Findings

Achieves competitive density estimation results.

Enables practical generation of 512x512 images.

Outperforms other non-autoregressive density models in quality and speed.

Abstract

PixelCNN achieves state-of-the-art results in density estimation for natural images. Although training is fast, inference is costly, requiring one network evaluation per pixel; O(N) for N pixels. This can be sped up by caching activations, but still involves generating each pixel sequentially. In this work, we propose a parallelized PixelCNN that allows more efficient inference by modeling certain pixel groups as conditionally independent. Our new PixelCNN model achieves competitive density estimation and orders of magnitude speedup - O(log N) sampling instead of O(N) - enabling the practical generation of 512x512 images. We evaluate the model on class-conditional image generation, text-to-image synthesis, and action-conditional video generation, showing that our model achieves the best results among non-pixel-autoregressive density models that allow efficient sampling.