Enhancing Sample Selection Against Label Noise by Cutting Mislabeled Easy Examples

TL;DR

This paper introduces Early Cutting, a method that improves learning with noisy labels by identifying and removing Mislabeled Easy Examples early in training, leading to better sample selection and model performance.

Contribution

The paper proposes a novel approach called Early Cutting that re-calibrates confident sample selection to effectively filter out harmful mislabeled easy examples during training.

Findings

Early Cutting reduces Mislabeled Easy Examples (MEEs) in training.

The method improves model accuracy on noisy datasets.

Experiments show consistent performance gains on CIFAR, WebVision, and ImageNet-1k.

Abstract

Sample selection is a prevalent approach in learning with noisy labels, aiming to identify confident samples for training. Although existing sample selection methods have achieved decent results by reducing the noise rate of the selected subset, they often overlook that not all mislabeled examples harm the model's performance equally. In this paper, we demonstrate that mislabeled examples correctly predicted by the model early in the training process are particularly harmful to model performance. We refer to these examples as Mislabeled Easy Examples (MEEs). To address this, we propose Early Cutting, which introduces a recalibration step that employs the model's later training state to re-select the confident subset identified early in training, thereby avoiding misleading confidence from early learning and effectively filtering out MEEs. Experiments on the CIFAR, WebVision, and full ImageNet-1k datasets demonstrate that our method effectively improves sample selection and model performance by reducing MEEs.