Learning from Noisy Labels with Deep Neural Networks: A Survey

TL;DR

This survey comprehensively reviews methods for training deep neural networks with noisy labels, analyzing their approaches, evaluation metrics, and datasets, to guide future research in robust learning.

Contribution

It provides a systematic categorization and comparison of 62 state-of-the-art robust training methods for noisy labels in deep learning.

Findings

Classification of methods into five groups based on methodology

Comparison of six evaluation properties for robustness

Analysis of noise rate estimation techniques

Abstract

Deep learning has achieved remarkable success in numerous domains with help from large amounts of big data. However, the quality of data labels is a concern because of the lack of high-quality labels in many real-world scenarios. As noisy labels severely degrade the generalization performance of deep neural networks, learning from noisy labels (robust training) is becoming an important task in modern deep learning applications. In this survey, we first describe the problem of learning with label noise from a supervised learning perspective. Next, we provide a comprehensive review of 62 state-of-the-art robust training methods, all of which are categorized into five groups according to their methodological difference, followed by a systematic comparison of six properties used to evaluate their superiority. Subsequently, we perform an in-depth analysis of noise rate estimation and summarize the typically used evaluation methodology, including public noisy datasets and evaluation metrics. Finally, we present several promising research directions that can serve as a guideline for future studies. All the contents will be available at https://github.com/songhwanjun/Awesome-Noisy-Labels.