Learning to Compose Domain-Specific Transformations for Data Augmentation

TL;DR

This paper introduces a method to automatically learn and compose data transformations for augmentation using a generative adversarial approach, improving model performance across image and text tasks.

Contribution

It presents a novel approach to automate the construction of complex data augmentation transformations using a generative sequence model trained adversarially.

Findings

Achieved 4.0 accuracy point improvement on CIFAR-10

Gained 1.4 F1 point improvement on ACE relation extraction

Improved 3.4 accuracy points with domain-specific transformations on medical imaging

Abstract

Data augmentation is a ubiquitous technique for increasing the size of labeled training sets by leveraging task-specific data transformations that preserve class labels. While it is often easy for domain experts to specify individual transformations, constructing and tuning the more sophisticated compositions typically needed to achieve state-of-the-art results is a time-consuming manual task in practice. We propose a method for automating this process by learning a generative sequence model over user-specified transformation functions using a generative adversarial approach. Our method can make use of arbitrary, non-deterministic transformation functions, is robust to misspecified user input, and is trained on unlabeled data. The learned transformation model can then be used to perform data augmentation for any end discriminative model. In our experiments, we show the efficacy of our approach on both image and text datasets, achieving improvements of 4.0 accuracy points on CIFAR-10, 1.4 F1 points on the ACE relation extraction task, and 3.4 accuracy points when using domain-specific transformation operations on a medical imaging dataset as compared to standard heuristic augmentation approaches.