• Home
  • Search
  • A Noisy Sample Selection Framework Based on a Mixup Loss and Recalibration Strategy
  • Cite Icon4
  • https://doi.org/10.3390/math12152389Copy DOI Icon

A Noisy Sample Selection Framework Based on a Mixup Loss and Recalibration Strategy

Show More
  • Abstract
  • Highlights & Summary
  • PDF
  • Literature Map
  • References
  • Citations
  • Similar Papers
Abstract

Deep neural networks (DNNs) have achieved breakthrough progress in various fields, largely owing to the support of large-scale datasets with manually annotated labels. However, obtaining such datasets is costly and time-consuming, making high-quality annotation a challenging task. In this work, we propose an improved noisy sample selection method, termed “sample selection framework”, based on a mixup loss and recalibration strategy (SMR). This framework enhances the robustness and generalization abilities of models. First, we introduce a robust mixup loss function to pre-train two models with identical structures separately. This approach avoids additional hyperparameter adjustments and reduces the need for prior knowledge of noise types. Additionally, we use a Gaussian Mixture Model (GMM) to divide the entire training set into labeled and unlabeled subsets, followed by robust training using semi-supervised learning (SSL) techniques. Furthermore, we propose a recalibration strategy based on cross-entropy (CE) loss to prevent the models from converging to local optima during the SSL process, thus further improving performance. Ablation experiments on CIFAR-10 with 50% symmetric noise and 40% asymmetric noise demonstrate that the two modules introduced in this paper improve the accuracy of the baseline (i.e., DivideMix) by 1.5% and 0.5%, respectively. Moreover, the experimental results on multiple benchmark datasets demonstrate that our proposed method effectively mitigates the impact of noisy labels and significantly enhances the performance of DNNs on noisy datasets. For instance, on the WebVision dataset, our method improves the top-1 accuracy by 0.7% and 2.4% compared to the baseline method.

Loading PDF

Similar Papers
  • Research Article
  • Citations2

BIOMETRIC HUMAN AUTHENTICATION SYSTEM THROUGH SPEECH USING DEEP NEURAL NETWORKS (DNN)

  • Oct 15, 2020
  • THE BULLETIN
  • O Mamyrbayev +4
  • Research Article
  • Citations61

Semi-Supervised Contrastive Learning With Similarity Co-Calibration

  • Jan 01, 2023
  • IEEE Transactions on Multimedia
  • Yuhang Zhang +5
  • Conference Article
  • Citations6

Novel Inter Mixture Weighted GMM Posteriorgram for DNN and GAN-based Voice Conversion

  • Nov 01, 2018
  • Nirmesh J Shah +3
  • Conference Article
  • Citations31

GMM-free DNN acoustic model training

  • May 01, 2014
  • Andrew Senior +3
  • PDF
  • Research Article
  • Citations29

Target Detection Network for SAR Images Based on Semi-Supervised Learning and Attention Mechanism

  • Jul 08, 2021
  • Remote Sensing
  • Di Wei +3
  • Conference Article
  • Citations110

Adaptive Consistency Regularization for Semi-Supervised Transfer Learning

  • Jun 01, 2021
  • Abulikemu Abuduweili +4
  • Research Article
  • Citations235

A stochastic frontier model with correction for sample selection

  • Nov 28, 2009
  • Journal of Productivity Analysis
  • William Greene
  • Conference Article
  • Citations3

CLAIM: An Enhanced Machine Learning Technique for Discrepancy Report Analysis

  • Jan 01, 2020
  • Myron Hecht +2
  • Research Article
  • Citations2

Mutual calibration training: Training deep neural networks with noisy labels using dual-models

  • Sep 13, 2021
  • Computer Vision and Image Understanding
  • Rui Liu +3
  • Conference Article
  • Citations29

Semi-supervised learning for semantic video retrieval

  • Jul 09, 2007
  • Ralph Ewerth +1
  • Research Article
  • Citations57

A Boosted Co-Training Algorithm for Human Action Recognition

  • Sep 01, 2011
  • IEEE Transactions on Circuits and Systems for Video Technology
  • Chang Liu +1
  • Book Chapter

Enhanced Non-linear Features for On-line Handwriting Recognition Using Deep Learning

  • Jan 01, 2014
  • Qing Zhang +3
  • Research Article
  • Citations86

Study of Senone-Based Deep Neural Network Approaches for Spoken Language Recognition

  • Jan 01, 2016
  • IEEE/ACM Transactions on Audio, Speech, and Language Processing
  • Luciana Ferrer +3
  • Conference Article
  • Citations100

Improving speaker recognition performance in the domain adaptation challenge using deep neural networks

  • Dec 01, 2014
  • Daniel Garcia-Romero +3
  • Conference Article
  • Citations2

Syllable-based Speech Recognition for a Very Low-Resource Language, Chaha

  • Dec 20, 2019
  • Tessfu Geteye Fantaye +2
Cactus Communications logo

Copyright 2026 Cactus Communications. All rights reserved.