Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labels

TL;DR

This paper introduces Co-teaching, a novel training paradigm where two neural networks teach each other to effectively learn from datasets with extremely noisy labels, significantly improving robustness over existing methods.

Contribution

The paper proposes Co-teaching, a new deep learning approach that trains two networks simultaneously to identify and focus on clean data, reducing the impact of noisy labels.

Findings

Co-teaching outperforms state-of-the-art methods on noisy MNIST, CIFAR-10, and CIFAR-100 datasets.

The method effectively prevents neural networks from memorizing noisy labels.

Empirical results demonstrate enhanced robustness of deep models trained with Co-teaching.

Abstract

Deep learning with noisy labels is practically challenging, as the capacity of deep models is so high that they can totally memorize these noisy labels sooner or later during training. Nonetheless, recent studies on the memorization effects of deep neural networks show that they would first memorize training data of clean labels and then those of noisy labels. Therefore in this paper, we propose a new deep learning paradigm called Co-teaching for combating with noisy labels. Namely, we train two deep neural networks simultaneously, and let them teach each other given every mini-batch: firstly, each network feeds forward all data and selects some data of possibly clean labels; secondly, two networks communicate with each other what data in this mini-batch should be used for training; finally, each network back propagates the data selected by its peer network and updates itself. Empirical results on noisy versions of MNIST, CIFAR-10 and CIFAR-100 demonstrate that Co-teaching is much superior to the state-of-the-art methods in the robustness of trained deep models.