Balanced Training for Sparse GANs

TL;DR

This paper introduces a new balanced dynamic sparse training method for GANs that maintains performance while reducing computational costs by controlling the balance between generator and discriminator.

Contribution

It proposes the balance ratio metric and the ADAPT method to effectively manage sparsity in GAN training, improving efficiency without sacrificing quality.

Findings

Effective reduction in training costs on multiple datasets

Maintains high-quality generated samples with sparse networks

Demonstrates improved balance between generator and discriminator

Abstract

Over the past few years, there has been growing interest in developing larger and deeper neural networks, including deep generative models like generative adversarial networks (GANs). However, GANs typically come with high computational complexity, leading researchers to explore methods for reducing the training and inference costs. One such approach gaining popularity in supervised learning is dynamic sparse training (DST), which maintains good performance while enjoying excellent training efficiency. Despite its potential benefits, applying DST to GANs presents challenges due to the adversarial nature of the training process. In this paper, we propose a novel metric called the balance ratio (BR) to study the balance between the sparse generator and discriminator. We also introduce a new method called balanced dynamic sparse training (ADAPT), which seeks to control the BR during GAN training to achieve a good trade-off between performance and computational cost. Our proposed method shows promising results on multiple datasets, demonstrating its effectiveness.