Latent Space is Feature Space: Regularization Term for GANs Training on Limited Dataset
This addresses the issue of data-hungry GANs for researchers and practitioners working with small datasets, though it is an incremental improvement over existing methods.
The paper tackles the problem of mode collapse in GANs when training on limited datasets by proposing LFM, an additional structure and loss function that maximizes feature diversity in the latent space, resulting in improved Frechet Inception Distance (FID) on the CelebA dataset.
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.