Graph-Attention Network with Adversarial Domain Alignment for Robust Cross-Domain Facial Expression Recognition

Razieh Ghaedi, AmirReza BabaAhmadi, Reyer Zwiggelaar, Xinqi Fan, Nashid Alam
Proceedings of the 17th Asian Conference on Machine Learning, PMLR 304:463-478, 2025.

Abstract

Cross-domain facial expression recognition (CD-FER) remains difficult due to severe domain shift between training and deployment data. We propose Graph-Attention Network with Adversarial Domain Alignment (GAT-ADA), a hybrid framework that couples a ResNet-50 as backbone with a batch-level Graph Attention Network (GAT) to model inter-sample relations under shift. Each mini-batch is cast as a sparse ring graph so that attention aggregates cross-sample cues that are informative for adaptation. To align distributions, GAT-ADA combines adversarial learning via a Gradient Reversal Layer (GRL) with statistical alignment using CORAL and MMD. GAT-ADA is evaluated under a standard unsupervised domain adaptation protocol: training on one labeled source (RAF-DB) and adapting to multiple unlabeled targets (CK+, JAFFE, SFEW 2.0, FER2013, and ExpW). GAT-ADA attains 74.39% mean cross-domain accuracy. On RAF-DB to FER2013, it reaches 98.0% accuracy, corresponding to a 36-point improvement over the best baseline we re-implemented with the same backbone and preprocessing.

Cite this Paper

BibTeX
@InProceedings{pmlr-v304-ghaedi25a, title = {Graph-Attention Network with Adversarial Domain Alignment for Robust Cross-Domain Facial Expression Recognition}, author = {Ghaedi, Razieh and BabaAhmadi, AmirReza and Zwiggelaar, Reyer and Fan, Xinqi and Alam, Nashid}, booktitle = {Proceedings of the 17th Asian Conference on Machine Learning}, pages = {463--478}, year = {2025}, editor = {Lee, Hung-yi and Liu, Tongliang}, volume = {304}, series = {Proceedings of Machine Learning Research}, month = {09--12 Dec}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v304/main/assets/ghaedi25a/ghaedi25a.pdf}, url = {https://proceedings.mlr.press/v304/ghaedi25a.html}, abstract = {Cross-domain facial expression recognition (CD-FER) remains difficult due to severe domain shift between training and deployment data. We propose Graph-Attention Network with Adversarial Domain Alignment (GAT-ADA), a hybrid framework that couples a ResNet-50 as backbone with a batch-level Graph Attention Network (GAT) to model inter-sample relations under shift. Each mini-batch is cast as a sparse ring graph so that attention aggregates cross-sample cues that are informative for adaptation. To align distributions, GAT-ADA combines adversarial learning via a Gradient Reversal Layer (GRL) with statistical alignment using CORAL and MMD. GAT-ADA is evaluated under a standard unsupervised domain adaptation protocol: training on one labeled source (RAF-DB) and adapting to multiple unlabeled targets (CK+, JAFFE, SFEW 2.0, FER2013, and ExpW). GAT-ADA attains 74.39% mean cross-domain accuracy. On RAF-DB to FER2013, it reaches 98.0% accuracy, corresponding to a 36-point improvement over the best baseline we re-implemented with the same backbone and preprocessing.} }
Endnote
%0 Conference Paper %T Graph-Attention Network with Adversarial Domain Alignment for Robust Cross-Domain Facial Expression Recognition %A Razieh Ghaedi %A AmirReza BabaAhmadi %A Reyer Zwiggelaar %A Xinqi Fan %A Nashid Alam %B Proceedings of the 17th Asian Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Hung-yi Lee %E Tongliang Liu %F pmlr-v304-ghaedi25a %I PMLR %P 463--478 %U https://proceedings.mlr.press/v304/ghaedi25a.html %V 304 %X Cross-domain facial expression recognition (CD-FER) remains difficult due to severe domain shift between training and deployment data. We propose Graph-Attention Network with Adversarial Domain Alignment (GAT-ADA), a hybrid framework that couples a ResNet-50 as backbone with a batch-level Graph Attention Network (GAT) to model inter-sample relations under shift. Each mini-batch is cast as a sparse ring graph so that attention aggregates cross-sample cues that are informative for adaptation. To align distributions, GAT-ADA combines adversarial learning via a Gradient Reversal Layer (GRL) with statistical alignment using CORAL and MMD. GAT-ADA is evaluated under a standard unsupervised domain adaptation protocol: training on one labeled source (RAF-DB) and adapting to multiple unlabeled targets (CK+, JAFFE, SFEW 2.0, FER2013, and ExpW). GAT-ADA attains 74.39% mean cross-domain accuracy. On RAF-DB to FER2013, it reaches 98.0% accuracy, corresponding to a 36-point improvement over the best baseline we re-implemented with the same backbone and preprocessing.
APA
Ghaedi, R., BabaAhmadi, A., Zwiggelaar, R., Fan, X. & Alam, N.. (2025). Graph-Attention Network with Adversarial Domain Alignment for Robust Cross-Domain Facial Expression Recognition. Proceedings of the 17th Asian Conference on Machine Learning, in Proceedings of Machine Learning Research 304:463-478 Available from https://proceedings.mlr.press/v304/ghaedi25a.html.

Related Material