Work

GTP-Net: A Graph–Transformer Framework for Cross-Subject EEG-Based Emotion Recognition

Year: 2025
Type: conference-paper
Abstract: EEG-based emotion recognition is crucial for affective brain–computer interfaces but remains challenging due to complex spatiotemporal dynamics. To address these issues, we propose GTP-Net, a graph–Transformer framework that combines graph attention for capturing inter-channel topology, temporal convolution for local dynamics, and a Transformer encoder with positional embeddings for long-range dependencies. Experiments on the SEED dataset demonstrate that GTP-Net achieves state-of-the-art accuracy and F1-score, outperforming competitive baselines and validating its effectiveness for reliable cross-subject EEG emotion recognition.
Authors: Junjie Li, Xu Chi, Dongsheng Zhu, Yixin Chen
Language: English
Open Access status: diamond
OpenAlex ID: W4416705192
FWCI:
Cites: 4
Cited by:
Related to: 10