Learning Energy-Based Models as Generative ConvNets via Multi-grid Modeling and Sampling

TL;DR

This paper introduces a multi-grid approach for training energy-based generative ConvNet models of images, improving synthesis quality by multi-scale MCMC sampling from coarse to fine levels.

Contribution

It presents a novel multi-grid learning algorithm that enhances energy-based ConvNet training by multi-scale sampling and outperforms previous contrastive divergence methods.

Findings

The multi-grid method effectively learns realistic image models.

It outperforms contrastive divergence and persistent CD in experiments.

Synthesized images are of higher quality and more diverse.

Abstract

This paper proposes a multi-grid method for learning energy-based generative ConvNet models of images. For each grid, we learn an energy-based probabilistic model where the energy function is defined by a bottom-up convolutional neural network (ConvNet or CNN). Learning such a model requires generating synthesized examples from the model. Within each iteration of our learning algorithm, for each observed training image, we generate synthesized images at multiple grids by initializing the finite-step MCMC sampling from a minimal 1 x 1 version of the training image. The synthesized image at each subsequent grid is obtained by a finite-step MCMC initialized from the synthesized image generated at the previous coarser grid. After obtaining the synthesized examples, the parameters of the models at multiple grids are updated separately and simultaneously based on the differences between synthesized and observed examples. We show that this multi-grid method can learn realistic energy-based generative ConvNet models, and it outperforms the original contrastive divergence (CD) and persistent CD.