TL;DR
RandAugment introduces a simplified, efficient automated data augmentation method that eliminates the need for a separate search phase, enabling easy application across various tasks and datasets with improved accuracy.
Contribution
It proposes RandAugment, a data augmentation approach with a reduced search space that can be trained directly on target tasks without proxy models or search phases.
Findings
Achieves 85.0% accuracy on ImageNet, surpassing previous methods.
Improves object detection mAP by 1.0-1.3% over baseline.
Matches or exceeds state-of-the-art results on multiple datasets.
Abstract
Recent work has shown that data augmentation has the potential to significantly improve the generalization of deep learning models. Recently, automated augmentation strategies have led to state-of-the-art results in image classification and object detection. While these strategies were optimized for improving validation accuracy, they also led to state-of-the-art results in semi-supervised learning and improved robustness to common corruptions of images. An obstacle to a large-scale adoption of these methods is a separate search phase which increases the training complexity and may substantially increase the computational cost. Additionally, due to the separate search phase, these approaches are unable to adjust the regularization strength based on model or dataset size. Automated augmentation policies are often found by training small models on small datasets and subsequently applied…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Code & Models
Videos
Taxonomy
MethodsHow to Speak to an United Airlines Live Support Agent: 10 Contact Methods Explained · Tanh Activation · Depthwise Convolution · Pointwise Convolution · Depthwise Separable Convolution · Residual Connection · Sigmoid Activation · Max Pooling · Long Short-Term Memory · Focal Loss
