# Adversarial Learning of General Transformations for Data Augmentation

arXiv: https://arxiv.org/abs/1909.09801 · HTML: https://mlsanity.com/paper/1909.09801/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 21 Sep 2019.

## Built on by (most cited first, 2 of 2)

- [Exploring Kolmogorov-Arnold Network Expansions in Vision Transformers for Mitigating Catastrophic Forgetting in Continual Learning](https://mlsanity.com/paper/2507.04020/) (2025)
- [Image Shape Manipulation from a Single Augmented Training Sample](https://mlsanity.com/paper/2109.06151/) (2021)

## Abstract (arXiv)

Data augmentation (DA) is fundamental against overfitting in large convolutional neural networks, especially with a limited training dataset. In images, DA is usually based on heuristic transformations, like geometric or color transformations. Instead of using predefined transformations, our work learns data augmentation directly from the training data by learning to transform images with an encoder-decoder architecture combined with a spatial transformer network. The transformed images still belong to the same class but are new, more complex samples for the classifier. Our experiments show that our approach is better than previous generative data augmentation methods, and comparable to predefined transformation methods when training an image classifier.