Instance Selection for GANs
arXiv:2007.15255
Abstract
Recent advances in Generative Adversarial Networks (GANs) have led to their widespread adoption for the purposes of generating high quality synthetic imagery. While capable of generating photo-realistic images, these models often produce unrealistic samples which fall outside of the data manifold. Several recently proposed techniques attempt to avoid spurious samples, either by rejecting them after generation, or by truncating the model's latent space. While effective, these methods are inefficient, as a large fraction of training time and model capacity are dedicated towards samples that will ultimately go unused. In this work we propose a novel approach to improve sample quality: altering the training dataset via instance selection before model training has taken place. By refining the empirical data distribution before training, we redirect model capacity towards high-density regions, which ultimately improves sample fidelity, lowers model capacity requirements, and significantly reduces training time. Code is available at https://github.com/uoguelph-mlrg/instance_selection_for_gans.
arXiv commentAccepted to NeurIPS 2020
References in corpus (14)
- Scikit-learn: Machine Learning in Python
- UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction
- GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
- Large Scale GAN Training for High Fidelity Natural Image Synthesis
- Self-Attention Generative Adversarial Networks
- Unsupervised Learning of Visual Features by Contrasting Cluster Assignments
- Which Training Methods for GANs do actually Converge?
- Classification Accuracy Score for Conditional Generative Models
- LOGAN: Latent Optimisation for Generative Adversarial Networks
- Your GAN is Secretly an Energy-based Model and You Should use Discriminator Driven Latent Sampling
- Megapixel Size Image Creation using Generative Adversarial Networks
- Small-GAN: Speeding Up GAN Training Using Core-sets
- Feature Quantization Improves GAN Training
- Subsampling Generative Adversarial Networks: Density Ratio Estimation in Feature Space with Softplus Loss