Early-Learning Regularization Prevents Memorization of Noisy Labels

TL;DR

This paper introduces a regularization-based method that leverages early learning phenomena in deep neural networks to prevent memorization of noisy labels, improving robustness in noisy classification tasks.

Contribution

The paper provides a theoretical understanding of early learning and memorization in high-dimensional models and proposes a novel regularization approach exploiting early learning for noise robustness.

Findings

Achieves robustness to noisy labels on standard benchmarks.

Provides theoretical insights into early learning and memorization.

Comparable performance to state-of-the-art methods on real-world datasets.

Abstract

We propose a novel framework to perform classification via deep learning in the presence of noisy annotations. When trained on noisy labels, deep neural networks have been observed to first fit the training data with clean labels during an "early learning" phase, before eventually memorizing the examples with false labels. We prove that early learning and memorization are fundamental phenomena in high-dimensional classification tasks, even in simple linear models, and give a theoretical explanation in this setting. Motivated by these findings, we develop a new technique for noisy classification tasks, which exploits the progress of the early learning phase. In contrast with existing approaches, which use the model output during early learning to detect the examples with clean labels, and either ignore or attempt to correct the false labels, we take a different route and instead capitalize on early learning via regularization. There are two key elements to our approach. First, we leverage semi-supervised learning techniques to produce target probabilities based on the model outputs. Second, we design a regularization term that steers the model towards these targets, implicitly preventing memorization of the false labels. The resulting framework is shown to provide robustness to noisy annotations on several standard benchmarks and real-world datasets, where it achieves results comparable to the state of the art.