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

Razieh Ghaedi, AmirReza BabaAhmadi, Reyer Zwiggelaar, Xinqi Fan, Nashid Alam · 2025

Read Graph-Attention Network with Adversarial Domain Alignment for Robust Cross-Domain Facial Expression Recognition by Razieh Ghaedi, AmirReza BabaAhmadi, Reyer Zwiggelaar, Xinqi Fan, Nashid Alam (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-00641.txt.

Description

Facial expression recognition often collapses when the people or cameras change between training and deployment. Razieh Ghaedi and co-authors propose GAT-ADA in 2025, a graph-attention network with adversarial domain alignment for cross-domain expression recognition. Severe domain shift is the central obstacle.

The framework pairs a ResNet-50 backbone with a batch-level graph attention network that models relations between samples under shift. Adversarial alignment pushes the model toward features that work in both domains. Robust expression recognition supports accessibility tools and human-computer interaction that must work for new users without retraining. Batch-level graphs let the model borrow strength from similar faces in the same batch. Adversarial alignment is the mechanism that forces domain-blind features.

Catalog record

CREATORRazieh Ghaedi, AmirReza BabaAhmadi, Reyer Zwiggelaar, Xinqi Fan, Nashid Alam
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE29 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-00641

Browse the full library · About the Tower of Babel