Symmetric Cross Entropy for Robust Learning with Noisy Labels
This addresses the challenge of robust learning in noisy-label scenarios for deep learning practitioners, offering an incremental improvement by enhancing existing methods.
The paper tackles the problem of training deep neural networks with noisy labels by identifying that cross entropy loss leads to overfitting on easy classes and underlearning on hard classes, and proposes Symmetric cross entropy Learning (SL) which combines cross entropy with a noise-tolerant reverse cross entropy to address both issues, achieving state-of-the-art performance on benchmark and real-world datasets.
Training accurate deep neural networks (DNNs) in the presence of noisy labels is an important and challenging task. Though a number of approaches have been proposed for learning with noisy labels, many open issues remain. In this paper, we show that DNN learning with Cross Entropy (CE) exhibits overfitting to noisy labels on some classes ("easy" classes), but more surprisingly, it also suffers from significant under learning on some other classes ("hard" classes). Intuitively, CE requires an extra term to facilitate learning of hard classes, and more importantly, this term should be noise tolerant, so as to avoid overfitting to noisy labels. Inspired by the symmetric KL-divergence, we propose the approach of \textbf{Symmetric cross entropy Learning} (SL), boosting CE symmetrically with a noise robust counterpart Reverse Cross Entropy (RCE). Our proposed SL approach simultaneously addresses both the under learning and overfitting problem of CE in the presence of noisy labels. We provide a theoretical analysis of SL and also empirically show, on a range of benchmark and real-world datasets, that SL outperforms state-of-the-art methods. We also show that SL can be easily incorporated into existing methods in order to further enhance their performance.