Abstract:In real applications, label noise has a great influence on data modeling. As one kind of label noise treatment method, noise filter has attracted extensive attention recently. The existing filters perform well in dealing with label noise completely at random (NCAR) but poorly in dealing with the label noise in the form of clusters (LLC). Besides, the existing filters may over remove the samples located at the classification boundaries, thereby affecting the generalization performance of classifiers. To fill these gaps, we propose a general elevating framework for label noise filters. The core idea of the framework is to improve the filtering performance of the existing filters by sample reduction. Specifically, since most of the existing filters are based on classifier prediction, and the complexity of samples at the boundary will affect the prediction performance of classifiers. Therefore, reducing the complexity of the boundary sample is very helpful to improve the performance of the filters. To this end, we propose a sample reduction method, which can not only reduce the complexity of the sample at the boundary but also convert LLC to NCAR, to get some representative samples. Next, the filter based on classifier prediction is employed to recognize the noisy representative samples. Finally, the noisy labeled samples in the given data set are found according to the identified noisy representatives. Furthermore, through empirical analysis, we found that compared with some classical metrics for evaluating the performance of noise filters, classification accuracy is more suitable to measure the performance of filters. Exhaustive experiments testify the validity of the framework, and the experimental results demonstrate that the performance of our framework is especially outstanding for LLC treatment.

Which is More Effective in Label Noise Cleaning, Correction or Filtering?

OT Cleaner: Label Correction As Optimal Transport

FGCM: Noisy Label Learning via Fine-Grained Confidence Modeling

Noise is the Fatal Poison: A Noise-aware Network for Noisy Dataset Classification

Learning from Noisy Labels with Decoupled Meta Label Purifier

A general elevating framework for label noise filters

Reliable Label Correction is a Good Booster When Learning with Extremely Noisy Labels.

Error-Bounded Correction of Noisy Labels

Uncertainty-guided label correction with wavelet-transformed discriminative representation enhancement

Two Wrongs Don't Make a Right: Combating Confirmation Bias in Learning with Label Noise.

Three-way Decision-Based Noise Correction for Crowdsourcing

Improving Crowdsourced Label Quality Using Noise Correction.

An joint end-to-end framework for learning with noisy labels

Synergistic Network Learning and Label Correction for Noise-robust Image Classification

Can We Treat Noisy Labels as Accurate?

Cleansing Noisy Data Streams

Noise cleaning for nonuniform ordinal labels based on inter-class distance

Learning to Purify Noisy Labels Via Meta Soft Label Corrector.

GNN Cleaner: Label Cleaner for Graph Structured Data

Label Noise: Correcting the Forward-Correction

Meta Label Correction for Noisy Label Learning