Machine Learning Research
GANs for Smaller Data: Training GANs on small data without overfitting
Trained on a small dataset, generative adversarial networks (GANs) tend to generate either replicas of the training data or noisy output. A new method spurs them to produce satisfying variations.