Towards harnessing feature embedding for robust learning with noisy labels

The memorization effect of deep neural networks (DNNs) plays a pivotal role in recent label noise learning methods. To exploit this effect, the model prediction-based methods have been widely adopted, which aim to exploit the outputs of DNNs in the early stage of learning to correct noisy labels. Ho...

Full description

Saved in:
Bibliographic Details
Published in:Machine learning Vol. 111; no. 9; pp. 3181 - 3201
Main Authors: Zhang, Chuang, Shen, Li, Yang, Jian, Gong, Chen
Format: Journal Article
Language:English
Published: New York Springer US 01.09.2022
Springer Nature B.V
Subjects:
Artificial Intelligence
Artificial neural networks
Computer Science
Control
Datasets
Deep learning
Dilution
Effectiveness
Embedding
Experiments
Labels
Machine Learning
Mechatronics
Natural Language Processing (NLP)
Neural networks
Robotics
Robustness
Simulation and Modeling
Supervision
Training
Deep learning
Robust learning
Label noise
Classification
ISSN:0885-6125, 1573-0565
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The memorization effect of deep neural networks (DNNs) plays a pivotal role in recent label noise learning methods. To exploit this effect, the model prediction-based methods have been widely adopted, which aim to exploit the outputs of DNNs in the early stage of learning to correct noisy labels. However, we observe that the model will make mistakes during label prediction, resulting in unsatisfactory performance. By contrast, the produced features in the early stage of learning show better robustness. Inspired by this observation, in this paper, we propose a novel feature embedding-based method for deep learning with label noise, termed L ab E l N oise D ilution (LEND). To be specific, we first compute a similarity matrix based on current embedded features to capture the local structure of training data. Then, the noisy supervision signals carried by mislabeled data are overwhelmed by nearby correctly labeled ones ( i.e. , label noise dilution), of which the effectiveness is guaranteed by the inherent robustness of feature embedding. Finally, the training data with diluted labels are further used to train a robust classifier. Empirically, we conduct extensive experiments on both synthetic and real-world noisy datasets by comparing our LEND with several representative robust learning approaches. The results verify the effectiveness of our LEND.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0885-6125
1573-0565
DOI:10.1007/s10994-022-06197-6