Analyze the Robustness of Classifiers under Label Noise

TL;DR

This paper investigates how label noise affects classifier robustness, proposing an approach that combines adversarial machine learning and importance reweighting with CNNs to improve resilience against noisy labels in real-world data.

Contribution

It introduces a novel method integrating AML and importance reweighting with CNNs to enhance classifier robustness under label noise conditions.

Findings

Improved model resilience to label noise

Effective focus on influential training samples

Enhanced performance in noisy data scenarios

Abstract

This study explores the robustness of label noise classifiers, aiming to enhance model resilience against noisy data in complex real-world scenarios. Label noise in supervised learning, characterized by erroneous or imprecise labels, significantly impairs model performance. This research focuses on the increasingly pertinent issue of label noise's impact on practical applications. Addressing the prevalent challenge of inaccurate training data labels, we integrate adversarial machine learning (AML) and importance reweighting techniques. Our approach involves employing convolutional neural networks (CNN) as the foundational model, with an emphasis on parameter adjustment for individual training samples. This strategy is designed to heighten the model's focus on samples critically influencing performance.