Loss-Sensitive Generative Adversarial Networks on Lipschitz Densities

TL;DR

This paper introduces Loss-Sensitive GANs with Lipschitz regularization, improving generalization and performance in image generation and classification by leveraging Lipschitz density conditions and encompassing models like Wasserstein GAN.

Contribution

The paper proposes a novel LS-GAN framework with Lipschitz regularization, unifying various GAN models including Wasserstein GAN, and extends it to conditional settings for enhanced classification.

Findings

LS-GAN and GLS-GAN generate competitive images with low reconstruction error.

Lipschitz regularization improves generalization over classic GANs.

Conditional LS-GAN performs well on image classification tasks.

Abstract

In this paper, we present the Lipschitz regularization theory and algorithms for a novel Loss-Sensitive Generative Adversarial Network (LS-GAN). Specifically, it trains a loss function to distinguish between real and fake samples by designated margins, while learning a generator alternately to produce realistic samples by minimizing their losses. The LS-GAN further regularizes its loss function with a Lipschitz regularity condition on the density of real data, yielding a regularized model that can better generalize to produce new data from a reasonable number of training examples than the classic GAN. We will further present a Generalized LS-GAN (GLS-GAN) and show it contains a large family of regularized GAN models, including both LS-GAN and Wasserstein GAN, as its special cases. Compared with the other GAN models, we will conduct experiments to show both LS-GAN and GLS-GAN exhibit competitive ability in generating new images in terms of the Minimum Reconstruction Error (MRE) assessed on a separate test set. We further extend the LS-GAN to a conditional form for supervised and semi-supervised learning problems, and demonstrate its outstanding performance on image classification tasks.