Latent Space is Feature Space: Regularization Term for GANs Training on Limited Dataset

TL;DR

This paper introduces LFM, a novel regularization method for GANs that enhances feature diversity in the latent space, reducing mode collapse and improving image quality with limited data.

Contribution

The paper proposes a new regularization term called LFM that maximizes feature diversity in the latent space, improving GAN training on small datasets.

Findings

LFM improves FID scores on CelebA dataset.

LFM reduces mode collapse in GANs.

The method requires mild additional computational performance.

Abstract

Generative Adversarial Networks (GAN) is currently widely used as an unsupervised image generation method. Current state-of-the-art GANs can generate photorealistic images with high resolution. However, a large amount of data is required, or the model would prone to generate images with similar patterns (mode collapse) and bad quality. I proposed an additional structure and loss function for GANs called LFM, trained to maximize the feature diversity between the different dimensions of the latent space to avoid mode collapse without affecting the image quality. Orthogonal latent vector pairs are created, and feature vector pairs extracted by discriminator are examined by dot product, with which discriminator and generator are in a novel adversarial relationship. In experiments, this system has been built upon DCGAN and proved to have improvement on Frechet Inception Distance (FID) training from scratch on CelebA Dataset. This system requires mild extra performance and can work with data augmentation methods. The code is available on github.com/penway/LFM.