Data Augmentation

Using diffusion models for generating synthetic training data

And why you might want to do it in the first place