Towards Deeper Generative Architectures for GANs using Dense connections

TL;DR

This paper explores the integration of skip connections and dense layers into Fisher GANs, demonstrating improved image quality with minimal impact from the number of connections.

Contribution

It introduces the application of dense and skip connections in Fisher GANs, showing their effectiveness in enhancing image generation quality.

Findings

Networks with connections produce better images.

Number of connections has minimal effect.

Dense connections improve GAN performance.

Abstract

In this paper, we present the result of adopting skip connections and dense layers, previously used in image classification tasks, in the Fisher GAN implementation. We have experimented with different numbers of layers and inserting these connections in different sections of the network. Our findings suggests that networks implemented with the connections produce better images than the baseline, and the number of connections added has only slight effect on the result.