Channel-Wise Contrastive Learning for Learning with Noisy Labels

TL;DR

This paper introduces channel-wise contrastive learning (CWCL), a novel approach that enhances learning with noisy labels by extracting authentic features through contrastive learning across channels, leading to more robust classifiers.

Contribution

The paper proposes CWCL, a new contrastive learning method that isolates genuine label features from noise, improving noisy label learning performance.

Findings

CWCL outperforms existing methods on benchmark datasets.

CWCL effectively identifies cleanly labeled samples.

The approach yields more nuanced and resilient features.

Abstract

In real-world datasets, noisy labels are pervasive. The challenge of learning with noisy labels (LNL) is to train a classifier that discerns the actual classes from given instances. For this, the model must identify features indicative of the authentic labels. While research indicates that genuine label information is embedded in the learned features of even inaccurately labeled data, it's often intertwined with noise, complicating its direct application. Addressing this, we introduce channel-wise contrastive learning (CWCL). This method distinguishes authentic label information from noise by undertaking contrastive learning across diverse channels. Unlike conventional instance-wise contrastive learning (IWCL), CWCL tends to yield more nuanced and resilient features aligned with the authentic labels. Our strategy is twofold: firstly, using CWCL to extract pertinent features to identify cleanly labeled samples, and secondly, progressively fine-tuning using these samples. Evaluations on several benchmark datasets validate our method's superiority over existing approaches.