IGAN: Inferent and Generative Adversarial Networks

TL;DR

IGAN introduces a neural architecture that jointly learns generative and inference models, providing stable training and a bidirectional mapping between data and latent space, suitable for complex high-dimensional data.

Contribution

It extends GANs with an inference mechanism, enabling a stable, bidirectional mapping and self-organized latent space without sacrificing simplicity or quality.

Findings

Enhanced stability and convergence in GAN training.

Successful application to image datasets including SAR and optical imagery.

Enables unsupervised exploitation of an enlarged latent space.

Abstract

I present IGAN (Inferent Generative Adversarial Networks), a neural architecture that learns both a generative and an inference model on a complex high dimensional data distribution, i.e. a bidirectional mapping between data samples and a simpler low-dimensional latent space. It extends the traditional GAN framework with inference by rewriting the adversarial strategy in both the image and the latent space with an entangled game between data-latent encoded posteriors and priors. It brings a measurable stability and convergence to the classical GAN scheme, while keeping its generative quality and remaining simple and frugal in order to run on a lab PC. IGAN fosters the encoded latents to span the full prior space: this enables the exploitation of an enlarged and self-organised latent space in an unsupervised manner. An analysis of previously published articles sets the theoretical ground for the proposed algorithm. A qualitative demonstration of potential applications like self-supervision or multi-modal data translation is given on common image datasets including SAR and optical imagery.