FCC-GAN: A Fully Connected and Convolutional Net Architecture for GANs

TL;DR

This paper introduces FCC-GAN, a novel GAN architecture combining fully connected and convolutional layers, which outperforms traditional convolution-only GANs in learning speed and sample quality across multiple datasets.

Contribution

The paper proposes FCC-GAN, a new GAN architecture that integrates fully connected layers with convolutional layers, demonstrating improved performance and stability.

Findings

FCC-GAN learns faster than traditional architectures

FCC-GAN generates higher quality samples

Effective across multiple image datasets

Abstract

Generative Adversarial Networks (GANs) are a powerful class of generative models. Despite their successes, the most appropriate choice of a GAN network architecture is still not well understood. GAN models for image synthesis have adopted a deep convolutional network architecture, which eliminates or minimizes the use of fully connected and pooling layers in favor of convolution layers in the generator and discriminator of GANs. In this paper, we demonstrate that a convolution network architecture utilizing deep fully connected layers and pooling layers can be more effective than the traditional convolution-only architecture, and we propose FCC-GAN, a fully connected and convolutional GAN architecture. Models based on our FCC-GAN architecture learn both faster than the conventional architecture and also generate higher quality of samples. We demonstrate the effectiveness and stability of our approach across four popular image datasets.