Viet-Anh on Software Logo

What is: AutoAugment?

SourceAutoAugment: Learning Augmentation Policies from Data
Year2000
Data SourceCC BY-SA - https://paperswithcode.com

AutoAugment is an automated approach to find data augmentation policies from data. It formulates the problem of finding the best augmentation policy as a discrete search problem. It consists of two components: a search algorithm and a search space.

At a high level, the search algorithm (implemented as a controller RNN) samples a data augmentation policy SS, which has information about what image processing operation to use, the probability of using the operation in each batch, and the magnitude of the operation. The policy SS is used to train a neural network with a fixed architecture, whose validation accuracy RR is sent back to update the controller. Since RR is not differentiable, the controller will be updated by policy gradient methods.

The operations used are from PIL, a popular Python image library: all functions in PIL that accept an image as input and output an image. It additionally uses two other augmentation techniques: Cutout and SamplePairing. The operations searched over are ShearX/Y, TranslateX/Y, Rotate, AutoContrast, Invert, Equalize, Solarize, Posterize, Contrast, Color, Brightness, Sharpness, Cutout and Sample Pairing.