Cross-Domain Facial Expression Recognition Based on Transductive Deep Transfer Learning

In this paper, we proposed a novel end-to-end transductive deep transfer learning network (TDTLN) to deal with the challenging cross-domain expression recognition problem, in which both the source and target databases are utilized to jointly learn optimal nonlinear discriminative features so as to i...

Full description

Saved in:
Bibliographic Details
Main Authors: Keyu Yan, Wenming Zheng, Tong Zhang, Yuan Zong, Chuangao Tang, Cheng Lu, Zhen Cui
Format: Article
Language:English
Published: IEEE 2019-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/8786815/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849725239781490688
author Keyu Yan
Wenming Zheng
Tong Zhang
Yuan Zong
Chuangao Tang
Cheng Lu
Zhen Cui
author_facet Keyu Yan
Wenming Zheng
Tong Zhang
Yuan Zong
Chuangao Tang
Cheng Lu
Zhen Cui
author_sort Keyu Yan
collection DOAJ
description In this paper, we proposed a novel end-to-end transductive deep transfer learning network (TDTLN) to deal with the challenging cross-domain expression recognition problem, in which both the source and target databases are utilized to jointly learn optimal nonlinear discriminative features so as to improve the label prediction performance of the target data samples. As part of the network parameters, the labels of the target samples are also optimized when optimizing the parameters of TDTLN, such that the cross-entropy loss of source domain data and the regression loss of target domain data can be simultaneously calculated. Finally, to evaluate the recognition performance of the proposed TDTLN method, we conduct extensive cross-database experiments on four commonly used multi-view facial expression databases, namely the BU-3DEF, Multi-PIE, SFEW, and RAF database. The experimental results show that the proposed TDTLN method outperforms state-of-the-art methods.
format Article
id doaj-art-fd4f79b252a84dffae3d9499c0098889
institution DOAJ
issn 2169-3536
language English
publishDate 2019-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-fd4f79b252a84dffae3d9499c00988892025-08-20T03:10:31ZengIEEEIEEE Access2169-35362019-01-01710890610891510.1109/ACCESS.2019.29303598786815Cross-Domain Facial Expression Recognition Based on Transductive Deep Transfer LearningKeyu Yan0https://orcid.org/0000-0001-6838-7203Wenming Zheng1https://orcid.org/0000-0002-7764-5179Tong Zhang2Yuan Zong3Chuangao Tang4Cheng Lu5Zhen Cui6Key Laboratory of Child Development and Learning Science of Ministry of Education, School of Biological Science and Medical Engineering, Southeast University, Nanjing, ChinaKey Laboratory of Child Development and Learning Science of Ministry of Education, School of Biological Science and Medical Engineering, Southeast University, Nanjing, ChinaSchool of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, ChinaKey Laboratory of Child Development and Learning Science of Ministry of Education, School of Biological Science and Medical Engineering, Southeast University, Nanjing, ChinaKey Laboratory of Child Development and Learning Science of Ministry of Education, School of Biological Science and Medical Engineering, Southeast University, Nanjing, ChinaKey Laboratory of Child Development and Learning Science of Ministry of Education, School of Biological Science and Medical Engineering, Southeast University, Nanjing, ChinaSchool of Computer Science and Engineering, Nanjing University of Science and Technology, Nanjing, ChinaIn this paper, we proposed a novel end-to-end transductive deep transfer learning network (TDTLN) to deal with the challenging cross-domain expression recognition problem, in which both the source and target databases are utilized to jointly learn optimal nonlinear discriminative features so as to improve the label prediction performance of the target data samples. As part of the network parameters, the labels of the target samples are also optimized when optimizing the parameters of TDTLN, such that the cross-entropy loss of source domain data and the regression loss of target domain data can be simultaneously calculated. Finally, to evaluate the recognition performance of the proposed TDTLN method, we conduct extensive cross-database experiments on four commonly used multi-view facial expression databases, namely the BU-3DEF, Multi-PIE, SFEW, and RAF database. The experimental results show that the proposed TDTLN method outperforms state-of-the-art methods.https://ieeexplore.ieee.org/document/8786815/Cross-domain facial expression recognitiontransductive transfer learningVGGFace16-Net
spellingShingle Keyu Yan
Wenming Zheng
Tong Zhang
Yuan Zong
Chuangao Tang
Cheng Lu
Zhen Cui
Cross-Domain Facial Expression Recognition Based on Transductive Deep Transfer Learning
IEEE Access
Cross-domain facial expression recognition
transductive transfer learning
VGGFace16-Net
title Cross-Domain Facial Expression Recognition Based on Transductive Deep Transfer Learning
title_full Cross-Domain Facial Expression Recognition Based on Transductive Deep Transfer Learning
title_fullStr Cross-Domain Facial Expression Recognition Based on Transductive Deep Transfer Learning
title_full_unstemmed Cross-Domain Facial Expression Recognition Based on Transductive Deep Transfer Learning
title_short Cross-Domain Facial Expression Recognition Based on Transductive Deep Transfer Learning
title_sort cross domain facial expression recognition based on transductive deep transfer learning
topic Cross-domain facial expression recognition
transductive transfer learning
VGGFace16-Net
url https://ieeexplore.ieee.org/document/8786815/
work_keys_str_mv AT keyuyan crossdomainfacialexpressionrecognitionbasedontransductivedeeptransferlearning
AT wenmingzheng crossdomainfacialexpressionrecognitionbasedontransductivedeeptransferlearning
AT tongzhang crossdomainfacialexpressionrecognitionbasedontransductivedeeptransferlearning
AT yuanzong crossdomainfacialexpressionrecognitionbasedontransductivedeeptransferlearning
AT chuangaotang crossdomainfacialexpressionrecognitionbasedontransductivedeeptransferlearning
AT chenglu crossdomainfacialexpressionrecognitionbasedontransductivedeeptransferlearning
AT zhencui crossdomainfacialexpressionrecognitionbasedontransductivedeeptransferlearning