Which Strategies Matter for Noisy Label Classification? Insight into Loss and Uncertainty

TL;DR

This paper analyzes how loss and uncertainty strategies affect training with noisy labels, revealing their complementary roles, and proposes a new robust method that improves performance across various datasets and models.

Contribution

It provides analytical insights into loss and uncertainty dynamics during training and introduces a novel method combining both to handle noisy labels effectively.

Findings

Our method outperforms state-of-the-art approaches.

It is effective across different neural network architectures.

The approach works well on both synthetic and real-world datasets.

Abstract

Label noise is a critical factor that degrades the generalization performance of deep neural networks, thus leading to severe issues in real-world problems. Existing studies have employed strategies based on either loss or uncertainty to address noisy labels, and ironically some strategies contradict each other: emphasizing or discarding uncertain samples or concentrating on high or low loss samples. To elucidate how opposing strategies can enhance model performance and offer insights into training with noisy labels, we present analytical results on how loss and uncertainty values of samples change throughout the training process. From the in-depth analysis, we design a new robust training method that emphasizes clean and informative samples, while minimizing the influence of noise using both loss and uncertainty. We demonstrate the effectiveness of our method with extensive experiments on synthetic and real-world datasets for various deep learning models. The results show that our method significantly outperforms other state-of-the-art methods and can be used generally regardless of neural network architectures.