SelfAugment: Automatic Augmentation Policies for Self-Supervised Learning

TL;DR

SelfAugment introduces an automatic method to select data augmentation policies for self-supervised learning by using a self-supervised evaluation task, eliminating the need for labeled data and achieving results comparable to supervised approaches.

Contribution

The paper proposes SelfAugment, a novel algorithm that automatically finds effective augmentation policies for self-supervised learning without relying on labeled data.

Findings

Self-supervised evaluation correlates highly with supervised evaluation (rank > 0.94).

SelfAugment achieves augmentation policies comparable to supervised methods.

The approach is effective across various architectures and training settings.

Abstract

A common practice in unsupervised representation learning is to use labeled data to evaluate the quality of the learned representations. This supervised evaluation is then used to guide critical aspects of the training process such as selecting the data augmentation policy. However, guiding an unsupervised training process through supervised evaluations is not possible for real-world data that does not actually contain labels (which may be the case, for example, in privacy sensitive fields such as medical imaging). Therefore, in this work we show that evaluating the learned representations with a self-supervised image rotation task is highly correlated with a standard set of supervised evaluations (rank correlation $> 0.94$). We establish this correlation across hundreds of augmentation policies, training settings, and network architectures and provide an algorithm (SelfAugment) to automatically and efficiently select augmentation policies without using supervised evaluations. Despite not using any labeled data, the learned augmentation policies perform comparably with augmentation policies that were determined using exhaustive supervised evaluations.