The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
Penyesuaian domain tanpa pengawasan (DA) ialah masalah pembelajaran mesin yang mencabar kerana set latihan (sumber) dan ujian tidak berlabel (sasaran) kepunyaan domain yang berbeza dan kemudian mempunyai pengedaran ciri yang berbeza, yang baru-baru ini telah menarik perhatian meluas dalam pengecaman ekspresi mikro (MER). ). Walaupun beberapa kaedah DA tanpa pengawasan yang berprestasi baik telah dicadangkan, kaedah ini tidak dapat menyelesaikan dengan baik masalah DA tanpa pengawasan dalam MER, aka, MER merentas domain. Untuk menangani masalah yang begitu mencabar, dalam surat ini kami mencadangkan kaedah DA baru tanpa pengawasan yang dipanggil pemberat Tampalan Bersama dan Padanan Momen (JPMM). JPMM merapatkan set ciri ekspresi mikro sumber dan sasaran dengan meminimumkan perbezaan taburan kebarangkalian mereka dengan operasi pemadanan momen berbilang pesanan. Sementara itu, ia mengambil kesempatan daripada tompok muka yang menyumbang dengan pembelajaran berat supaya perwakilan ciri invarian domain yang melibatkan maklumat yang boleh dibezakan ekspresi mikro boleh dipelajari. Akhir sekali, kami menjalankan eksperimen yang meluas untuk menilai kaedah JPMM yang dicadangkan adalah lebih baik daripada kaedah DA tanpa pengawasan terkini dalam menangani MER silang domain.
Jie ZHU
Southeast University
Yuan ZONG
Southeast University
Hongli CHANG
Southeast University
Li ZHAO
Southeast University
Chuangao TANG
Southeast University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Salinan
Jie ZHU, Yuan ZONG, Hongli CHANG, Li ZHAO, Chuangao TANG, "Joint Patch Weighting and Moment Matching for Unsupervised Domain Adaptation in Micro-Expression Recognition" in IEICE TRANSACTIONS on Information,
vol. E105-D, no. 2, pp. 441-445, February 2022, doi: 10.1587/transinf.2021EDL8045.
Abstract: Unsupervised domain adaptation (DA) is a challenging machine learning problem since the labeled training (source) and unlabeled testing (target) sets belong to different domains and then have different feature distributions, which has recently attracted wide attention in micro-expression recognition (MER). Although some well-performing unsupervised DA methods have been proposed, these methods cannot well solve the problem of unsupervised DA in MER, a. k. a., cross-domain MER. To deal with such a challenging problem, in this letter we propose a novel unsupervised DA method called Joint Patch weighting and Moment Matching (JPMM). JPMM bridges the source and target micro-expression feature sets by minimizing their probability distribution divergence with a multi-order moment matching operation. Meanwhile, it takes advantage of the contributive facial patches by the weight learning such that a domain-invariant feature representation involving micro-expression distinguishable information can be learned. Finally, we carry out extensive experiments to evaluate the proposed JPMM method is superior to recent state-of-the-art unsupervised DA methods in dealing with cross-domain MER.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2021EDL8045/_p
Salinan
@ARTICLE{e105-d_2_441,
author={Jie ZHU, Yuan ZONG, Hongli CHANG, Li ZHAO, Chuangao TANG, },
journal={IEICE TRANSACTIONS on Information},
title={Joint Patch Weighting and Moment Matching for Unsupervised Domain Adaptation in Micro-Expression Recognition},
year={2022},
volume={E105-D},
number={2},
pages={441-445},
abstract={Unsupervised domain adaptation (DA) is a challenging machine learning problem since the labeled training (source) and unlabeled testing (target) sets belong to different domains and then have different feature distributions, which has recently attracted wide attention in micro-expression recognition (MER). Although some well-performing unsupervised DA methods have been proposed, these methods cannot well solve the problem of unsupervised DA in MER, a. k. a., cross-domain MER. To deal with such a challenging problem, in this letter we propose a novel unsupervised DA method called Joint Patch weighting and Moment Matching (JPMM). JPMM bridges the source and target micro-expression feature sets by minimizing their probability distribution divergence with a multi-order moment matching operation. Meanwhile, it takes advantage of the contributive facial patches by the weight learning such that a domain-invariant feature representation involving micro-expression distinguishable information can be learned. Finally, we carry out extensive experiments to evaluate the proposed JPMM method is superior to recent state-of-the-art unsupervised DA methods in dealing with cross-domain MER.},
keywords={},
doi={10.1587/transinf.2021EDL8045},
ISSN={1745-1361},
month={February},}
Salinan
TY - JOUR
TI - Joint Patch Weighting and Moment Matching for Unsupervised Domain Adaptation in Micro-Expression Recognition
T2 - IEICE TRANSACTIONS on Information
SP - 441
EP - 445
AU - Jie ZHU
AU - Yuan ZONG
AU - Hongli CHANG
AU - Li ZHAO
AU - Chuangao TANG
PY - 2022
DO - 10.1587/transinf.2021EDL8045
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E105-D
IS - 2
JA - IEICE TRANSACTIONS on Information
Y1 - February 2022
AB - Unsupervised domain adaptation (DA) is a challenging machine learning problem since the labeled training (source) and unlabeled testing (target) sets belong to different domains and then have different feature distributions, which has recently attracted wide attention in micro-expression recognition (MER). Although some well-performing unsupervised DA methods have been proposed, these methods cannot well solve the problem of unsupervised DA in MER, a. k. a., cross-domain MER. To deal with such a challenging problem, in this letter we propose a novel unsupervised DA method called Joint Patch weighting and Moment Matching (JPMM). JPMM bridges the source and target micro-expression feature sets by minimizing their probability distribution divergence with a multi-order moment matching operation. Meanwhile, it takes advantage of the contributive facial patches by the weight learning such that a domain-invariant feature representation involving micro-expression distinguishable information can be learned. Finally, we carry out extensive experiments to evaluate the proposed JPMM method is superior to recent state-of-the-art unsupervised DA methods in dealing with cross-domain MER.
ER -