Noise Homogenization via Multi-Channel Wavelet Filtering for High-Fidelity Sample Generation in GANs

TL;DR

This paper introduces WaveletGAN, a novel GAN architecture that employs multi-channel wavelet filtering to homogenize noise, resulting in higher fidelity sample generation demonstrated by superior FID scores on multiple datasets.

Contribution

The paper proposes a wavelet deconvolution layer in GANs to improve noise homogenization and sample quality, a novel approach not previously explored.

Findings

WaveletGAN achieves the lowest FID scores on Fashion-MNIST, KMNIST, and SVHN.

Wavelet filtering effectively homogenizes noise in GANs.

The method enhances sample fidelity compared to traditional GANs.

Abstract

In the generator of typical Generative Adversarial Networks (GANs), a noise is inputted to generate fake samples via a series of convolutional operations. However, current noise generation models merely relies on the information from the pixel space, which increases the difficulty to approach the target distribution. Fortunately, the long proven wavelet transformation is able to decompose multiple spectral information from the images. In this work, we propose a novel multi-channel wavelet-based filtering method for GANs, to cope with this problem. When embedding a wavelet deconvolution layer in the generator, the resultant GAN, called WaveletGAN, takes advantage of the wavelet deconvolution to learn a filtering with multiple channels, which can efficiently homogenize the generated noise via an averaging operation, so as to generate high-fidelity samples. We conducted benchmark experiments on the Fashion-MNIST, KMNIST and SVHN datasets through an open GAN benchmark tool. The results show that WaveletGAN has excellent performance in generating high-fidelity samples, thanks to the smallest FIDs obtained on these datasets.