Updating the generator in PPGN-h with gradients flowing through the encoder

TL;DR

This paper proposes an extension to the Plug & Play generative model by incorporating discriminators that evaluate the authenticity of latent codes, with experiments confirming its viability on MNIST.

Contribution

It introduces a novel discriminator-based approach to improve the Plug & Play model by assessing latent codes, enhancing generative quality.

Findings

Discriminators effectively distinguish real and fake latent codes.

The approach is viable and improves the generative process.

Validated on MNIST dataset.

Abstract

The Generative Adversarial Network framework has shown success in implicitly modeling data distributions and is able to generate realistic samples. Its architecture is comprised of a generator, which produces fake data that superficially seem to belong to the real data distribution, and a discriminator which is to distinguish fake from genuine samples. The Noiseless Joint Plug & Play model offers an extension to the framework by simultaneously training autoencoders. This model uses a pre-trained encoder as a feature extractor, feeding the generator with global information. Using the Plug & Play network as baseline, we design a new model by adding discriminators to the Plug & Play architecture. These additional discriminators are trained to discern real and fake latent codes, which are the output of the encoder using genuine and generated inputs, respectively. We proceed to investigate whether this approach is viable. Experiments conducted for the MNIST manifold show that this indeed is the case.