Self-training with Noisy Student improves ImageNet classification
This work addresses the challenge of enhancing model performance and robustness in computer vision, particularly for ImageNet classification, with incremental improvements over existing self-training and distillation methods.
The paper tackles the problem of improving ImageNet classification accuracy and robustness by introducing Noisy Student Training, a semi-supervised learning method that achieves 88.4% top-1 accuracy on ImageNet (2.0% better than previous state-of-the-art) and significantly improves performance on robustness test sets like ImageNet-A (from 61.0% to 83.7% top-1 accuracy).
We present Noisy Student Training, a semi-supervised learning approach that works well even when labeled data is abundant. Noisy Student Training achieves 88.4% top-1 accuracy on ImageNet, which is 2.0% better than the state-of-the-art model that requires 3.5B weakly labeled Instagram images. On robustness test sets, it improves ImageNet-A top-1 accuracy from 61.0% to 83.7%, reduces ImageNet-C mean corruption error from 45.7 to 28.3, and reduces ImageNet-P mean flip rate from 27.8 to 12.2. Noisy Student Training extends the idea of self-training and distillation with the use of equal-or-larger student models and noise added to the student during learning. On ImageNet, we first train an EfficientNet model on labeled images and use it as a teacher to generate pseudo labels for 300M unlabeled images. We then train a larger EfficientNet as a student model on the combination of labeled and pseudo labeled images. We iterate this process by putting back the student as the teacher. During the learning of the student, we inject noise such as dropout, stochastic depth, and data augmentation via RandAugment to the student so that the student generalizes better than the teacher. Models are available at https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet. Code is available at https://github.com/google-research/noisystudent.