Faster AutoAugment: Learning Augmentation Strategies using Backpropagation

TL;DR

Faster AutoAugment introduces a differentiable approach to learn data augmentation strategies efficiently, significantly reducing search time while maintaining high performance in image recognition tasks.

Contribution

It proposes a novel differentiable policy search method for data augmentation, enabling rapid discovery of effective strategies compared to traditional black-box search algorithms.

Findings

Achieves faster search times than previous AutoAugment methods.

Maintains comparable or better augmentation performance.

Introduces differentiable mechanisms for discrete transformation parameters.

Abstract

Data augmentation methods are indispensable heuristics to boost the performance of deep neural networks, especially in image recognition tasks. Recently, several studies have shown that augmentation strategies found by search algorithms outperform hand-made strategies. Such methods employ black-box search algorithms over image transformations with continuous or discrete parameters and require a long time to obtain better strategies. In this paper, we propose a differentiable policy search pipeline for data augmentation, which is much faster than previous methods. We introduce approximate gradients for several transformation operations with discrete parameters as well as the differentiable mechanism for selecting operations. As the objective of training, we minimize the distance between the distributions of augmented data and the original data, which can be differentiated. We show that our method, Faster AutoAugment, achieves significantly faster searching than prior work without a performance drop.