Unsupervised feature learning by augmenting single images

TL;DR

This paper proposes an unsupervised feature learning method using data augmentation on single images, where a CNN learns to discriminate between transformed patches, resulting in competitive performance on standard datasets.

Contribution

It introduces a novel unsupervised learning approach that leverages data augmentation as the core mechanism for feature learning from single images.

Findings

Achieves competitive classification results on STL-10, CIFAR-10, Caltech-101

Demonstrates effectiveness of augmentation-based surrogate classes for unsupervised learning

Shows simple augmentation-based training can rival more complex methods

Abstract

When deep learning is applied to visual object recognition, data augmentation is often used to generate additional training data without extra labeling cost. It helps to reduce overfitting and increase the performance of the algorithm. In this paper we investigate if it is possible to use data augmentation as the main component of an unsupervised feature learning architecture. To that end we sample a set of random image patches and declare each of them to be a separate single-image surrogate class. We then extend these trivial one-element classes by applying a variety of transformations to the initial 'seed' patches. Finally we train a convolutional neural network to discriminate between these surrogate classes. The feature representation learned by the network can then be used in various vision tasks. We find that this simple feature learning algorithm is surprisingly successful, achieving competitive classification results on several popular vision datasets (STL-10, CIFAR-10, Caltech-101).