Robust Temporal Ensembling for Learning with Noisy Labels

09/29/2021
by   Abel Brown, et al.
0

Successful training of deep neural networks with noisy labels is an essential capability as most real-world datasets contain some amount of mislabeled data. Left unmitigated, label noise can sharply degrade typical supervised learning approaches. In this paper, we present robust temporal ensembling (RTE), which combines robust loss with semi-supervised regularization methods to achieve noise-robust learning. We demonstrate that RTE achieves state-of-the-art performance across the CIFAR-10, CIFAR-100, ImageNet, WebVision, and Food-101N datasets, while forgoing the recent trend of label filtering and/or fixing. Finally, we show that RTE also retains competitive corruption robustness to unforeseen input noise using CIFAR-10-C, obtaining a mean corruption error (mCE) of 13.50 with standard methods on clean data.

READ FULL TEXT
research
03/06/2021

LongReMix: Robust Learning with High Confidence Samples in a Noisy Label Environment

Deep neural network models are robust to a limited amount of label noise...
research
10/04/2021

Consistency Regularization Can Improve Robustness to Label Noise

Consistency regularization is a commonly-used technique for semi-supervi...
research
05/31/2017

Toward Robustness against Label Noise in Training Deep Discriminative Neural Networks

Collecting large training datasets, annotated with high-quality labels, ...
research
12/03/2022

CrossSplit: Mitigating Label Noise Memorization through Data Splitting

We approach the problem of improving robustness of deep learning algorit...
research
05/28/2022

Deep Learning with Label Noise: A Hierarchical Approach

Deep neural networks are susceptible to label noise. Existing methods to...
research
02/04/2022

Learning with Neighbor Consistency for Noisy Labels

Recent advances in deep learning have relied on large, labelled datasets...
research
12/02/2020

SemiNLL: A Framework of Noisy-Label Learning by Semi-Supervised Learning

Deep learning with noisy labels is a challenging task. Recent prominent ...

Please sign up or login with your details

Forgot password? Click here to reset