When do GANs replicate? On the choice of dataset size
Authors:
- 1 October 2021
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 6681-6690
Abstract
Do GANs replicate training images? Previous studies have shown that GANs do not seem to replicate training data without significant change in the training procedure. This leads to a series of research on the exact condition needed for GANs to overfit to the training data. Although a number of factors has been theoretically or empirically identified, the effect of dataset size and complexity on GANs replication is still unknown. With empirical evidence from BigGAN and StyleGAN2, on datasets CelebA, Flower and LSUN-bedroom, we show that dataset size and its complexity play an important role in GANs replication and perceptual quality of the generated images. We further quantify this relationship, discovering that replication percentage decays exponentially with respect to dataset size and complexity, with a shared decaying factor across GAN-dataset combinations. Meanwhile, the perceptual image quality follows a U-shape trend w.r.t dataset size. This finding leads to a practical tool for one-shot estimation on minimal dataset size to prevent GAN replication which can be used to guide datasets construction and selection.Keywords
All Related Versions
This publication has 16 references indexed in Scilit:
- On the Intrinsic Dimensionality of Image Representations2019
- Detecting Overfitting of Deep Generative Networks via Latent Recovery2019
- A Style-Based Generator Architecture for Generative Adversarial Networks2019
- What Does It Mean to Learn in Deep Networks? And, How Does One Detect Adversarial Attacks?2019
- Generative adversarial networks and convolutional neural networks based weather classification model for day ahead short-term photovoltaic power forecastingEnergy Conversion and Management, 2019
- GAN-based synthetic medical image augmentation for increased CNN performance in liver lesion classificationNeurocomputing, 2018
- Effective data generation for imbalanced learning using conditional generative adversarial networksExpert Systems with Applications, 2018
- Deep Learning Face Attributes in the Wild2015
- Automated Flower Classification over a Large Number of Classes2008
- Communication in the Presence of NoiseProceedings of the IRE, 1949