Student's t-Generative Adversarial Networks

TL;DR

This paper introduces Student's t-Generative Adversarial Networks, which incorporate Student's t-distribution into the latent space to enhance diversity and performance in image generation with limited data.

Contribution

The paper proposes a novel GAN variant using Student's t-distribution for the latent space and a dual-task discriminator, improving diversity and quality of generated images.

Findings

Enhanced diversity in generated images

Better performance with limited training data

Mathematical proof of distribution transformation

Abstract

Generative Adversarial Networks (GANs) have a great performance in image generation, but they need a large scale of data to train the entire framework, and often result in nonsensical results. We propose a new method referring to conditional GAN, which equipments the latent noise with mixture of Student's t-distribution with attention mechanism in addition to class information. Student's t-distribution has long tails that can provide more diversity to the latent noise. Meanwhile, the discriminator in our model implements two tasks simultaneously, judging whether the images come from the true data distribution, and identifying the class of each generated images. The parameters of the mixture model can be learned along with those of GANs. Moreover, we mathematically prove that any multivariate Student's t-distribution can be obtained by a linear transformation of a normal multivariate Student's t-distribution. Experiments comparing the proposed method with typical GAN, DeliGAN and DCGAN indicate that, our method has a great performance on generating diverse and legible objects with limited data.