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. |
Year | DOI | Venue |
---|---|---|
2021 | 10.1109/CVPR46437.2021.00270 | 2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021 |
Keywords | DocType | ISSN |
Feature learning,Machine learning,Network architecture,Computer science,Medical imaging,Rank correlation,Correlation,Artificial intelligence,Image rotation,Labeled data,Self supervised learning | Conference | 1063-6919 |
Citations | PageRank | References |
0 | 0.34 | 13 |
Authors | ||
5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Reed Colorado J | 1 | 0 | 0.34 |
Sean Metzger | 2 | 0 | 0.68 |
Aravind S. Lakshminarayanan | 3 | 24 | 5.41 |
Trevor Darrell | 4 | 22413 | 1800.67 |
Kurt Keutzer | 5 | 18 | 4.86 |