Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Liu, Qiuhao, Li, Ling, Lu, Yao, Xuan, Qi, Zhu, Zhaowei, Wei, Jiaheng
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2509.11265
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915494212141056
author Liu, Qiuhao
Li, Ling
Lu, Yao
Xuan, Qi
Zhu, Zhaowei
Wei, Jiaheng
author_facet Liu, Qiuhao
Li, Ling
Lu, Yao
Xuan, Qi
Zhu, Zhaowei
Wei, Jiaheng
contents Deep neural networks tend to memorize noisy labels, severely degrading their generalization performance. Although Mixup has demonstrated effectiveness in improving generalization and robustness, existing Mixup-based methods typically perform indiscriminate mixing without principled guidance on sample selection and mixing strategy, inadvertently propagating noisy supervision. To overcome these limitations, we propose SelectMix, a confidence-guided mixing framework explicitly tailored for noisy labels. SelectMix first identifies potentially noisy or ambiguous samples through confidence based mismatch analysis using K-fold cross-validation, then selectively blends identified uncertain samples with confidently predicted peers from their potential classes. Furthermore, SelectMix employs soft labels derived from all classes involved in the mixing process, ensuring the labels accurately represent the composition of the mixed samples, thus aligning supervision signals closely with the actual mixed inputs. Through extensive theoretical analysis and empirical evaluations on multiple synthetic (MNIST, Fashion-MNIST, CIFAR-10, CIFAR-100) and real-world benchmark datasets (CIFAR-N, MNIST and Clothing1M), we demonstrate that SelectMix consistently outperforms strong baseline methods, validating its effectiveness and robustness in learning with noisy labels.
format Preprint
id arxiv_https___arxiv_org_abs_2509_11265
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SelectMix: Enhancing Label Noise Robustness through Targeted Sample Mixing
Liu, Qiuhao
Li, Ling
Lu, Yao
Xuan, Qi
Zhu, Zhaowei
Wei, Jiaheng
Machine Learning
Computer Vision and Pattern Recognition
Deep neural networks tend to memorize noisy labels, severely degrading their generalization performance. Although Mixup has demonstrated effectiveness in improving generalization and robustness, existing Mixup-based methods typically perform indiscriminate mixing without principled guidance on sample selection and mixing strategy, inadvertently propagating noisy supervision. To overcome these limitations, we propose SelectMix, a confidence-guided mixing framework explicitly tailored for noisy labels. SelectMix first identifies potentially noisy or ambiguous samples through confidence based mismatch analysis using K-fold cross-validation, then selectively blends identified uncertain samples with confidently predicted peers from their potential classes. Furthermore, SelectMix employs soft labels derived from all classes involved in the mixing process, ensuring the labels accurately represent the composition of the mixed samples, thus aligning supervision signals closely with the actual mixed inputs. Through extensive theoretical analysis and empirical evaluations on multiple synthetic (MNIST, Fashion-MNIST, CIFAR-10, CIFAR-100) and real-world benchmark datasets (CIFAR-N, MNIST and Clothing1M), we demonstrate that SelectMix consistently outperforms strong baseline methods, validating its effectiveness and robustness in learning with noisy labels.
title SelectMix: Enhancing Label Noise Robustness through Targeted Sample Mixing
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2509.11265