Self-Diagnosing GAN: Diagnosing Underrepresented Samples in Generative Adversarial Networks
This addresses the issue of sample diversity and quality in GANs for applications like data generation, but it is incremental as it builds on existing GAN training techniques.
The paper tackles the problem of GANs producing low-quality samples for underrepresented groups by proposing a method to diagnose and emphasize these samples during training, resulting in improved quality and diversity for minor groups across various datasets.
Despite remarkable performance in producing realistic samples, Generative Adversarial Networks (GANs) often produce low-quality samples near low-density regions of the data manifold, e.g., samples of minor groups. Many techniques have been developed to improve the quality of generated samples, either by post-processing generated samples or by pre-processing the empirical data distribution, but at the cost of reduced diversity. To promote diversity in sample generation without degrading the overall quality, we propose a simple yet effective method to diagnose and emphasize underrepresented samples during training of a GAN. The main idea is to use the statistics of the discrepancy between the data distribution and the model distribution at each data instance. Based on the observation that the underrepresented samples have a high average discrepancy or high variability in discrepancy, we propose a method to emphasize those samples during training of a GAN. Our experimental results demonstrate that the proposed method improves GAN performance on various datasets, and it is especially effective in improving the quality and diversity of sample generation for minor groups.