Circumventing Outliers of AutoAugment with Knowledge Distillation
arXiv:2003.11342
Abstract
AutoAugment has been a powerful algorithm that improves the accuracy of many vision tasks, yet it is sensitive to the operator space as well as hyper-parameters, and an improper setting may degenerate network optimization. This paper delves deep into the working mechanism, and reveals that AutoAugment may remove part of discriminative information from the training image and so insisting on the ground-truth label is no longer the best option. To relieve the inaccuracy of supervision, we make use of knowledge distillation that refers to the output of a teacher model to guide network training. Experiments are performed in standard image classification benchmarks, and demonstrate the effectiveness of our approach in suppressing noise of data augmentation and stabilizing training. Upon the cooperation of knowledge distillation and AutoAugment, we claim the new state-of-the-art on ImageNet classification with a top-1 accuracy of 85.8%.
arXiv comment18 pages, 4 figures
References in corpus (17)
- Distilling the Knowledge in a Neural Network
- Neural Architecture Search with Reinforcement Learning
- Improved Regularization of Convolutional Neural Networks with Cutout
- FitNets: Hints for Thin Deep Nets
- Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results
- DARTS: Differentiable Architecture Search
- Random Erasing Data Augmentation
- Large-Scale Evolution of Image Classifiers
- Billion-scale semi-supervised learning for image classification
- Shake-Shake regularization
- Self-training with Noisy Student improves ImageNet classification
- ShakeDrop Regularization for Deep Residual Learning
- Label Refinery: Improving ImageNet Classification through Label Progression
- Population Based Augmentation: Efficient Learning of Augmentation Policy Schedules
- Adversarial Examples Improve Image Recognition
- Data Augmentation Revisited: Rethinking the Distribution Gap between Clean and Augmented Data
- Fixing the train-test resolution discrepancy