6
2

Distilling Effective Supervision from Severe Label Noise

Abstract

Collecting large-scale data with clean labels for supervised training of neural networks is practically challenging. Although noisy labels are usually cheap to acquire, existing methods suffer a lot from label noise. This paper targets at the challenge of robust training at high label noise regimes. The key insight to achieve this goal is to wisely leverage a small trusted set to estimate exemplar weights and pseudo labels for noisy data in order to reuse them for supervised training. We present a holistic framework to train deep neural networks in a way that is highly invulnerable to label noise. Our method sets the new state of the art on various types of label noise and achieves excellent performance on large-scale datasets with real-world label noise. For instance, on CIFAR100 with a 40%40\% uniform noise ratio and only 10 trusted labeled data per class, our method achieves 80.2±0.3%80.2{\pm}0.3\% classification accuracy, where the error rate is only 1.4%1.4\% higher than a neural network trained without label noise. Moreover, increasing the noise ratio to 80%80\%, our method still maintains a high accuracy of 75.5±0.2%75.5{\pm}0.2\%, compared to the previous best accuracy 48.2%48.2\%. Source code available: https://github.com/google-research/google-research/tree/master/ieg

View on arXiv
Comments on this paper