[Paper Review] Graph Neural Networks in EEG-based Emotion Recognition: A Survey

Chenyu Liu, Deng, Yuqiu+11 more|ArXiv.org|Feb 2, 2024

EEG and Brain-Computer Interfaces11 citations

TL;DR

This survey comprehensively categorizes GNN approaches for EEG-based emotion recognition within a unified three-stage framework (node, edge, graph) and outlines future directions.

ABSTRACT

Compared to other modalities, EEG-based emotion recognition can intuitively respond to the emotional patterns in the human brain and, therefore, has become one of the most concerning tasks in the brain-computer interfaces field. Since dependencies within brain regions are closely related to emotion, a significant trend is to develop Graph Neural Networks (GNNs) for EEG-based emotion recognition. However, brain region dependencies in emotional EEG have physiological bases that distinguish GNNs in this field from those in other time series fields. Besides, there is neither a comprehensive review nor guidance for constructing GNNs in EEG-based emotion recognition. In the survey, our categorization reveals the commonalities and differences of existing approaches under a unified framework of graph construction. We analyze and categorize methods from three stages in the framework to provide clear guidance on constructing GNNs in EEG-based emotion recognition. In addition, we discuss several open challenges and future directions, such as Temporal full-connected graph and Graph condensation.

Motivation & Objective

• Provide a comprehensive and systematic review of existing GNNs in EEG-based emotion recognition.
• Propose a novel categorization of GNNs under a unified three-stage framework (node-level, edge-level, graph-level).
• Offer guidance for constructing GNNs for EEG-emotion tasks and highlight open challenges and future directions.

Proposed method

• Frame EEG-based emotion recognition as f(G) mapping from graph G to emotion label Y.
• Categorize methods by three construction stages: node-level feature choice, edge computation, and graph structure.
• Introduce node-feature categories (Univariate vs Hybrid) and edge types (Model-independent vs Model-dependent).
• Describe graph-level structures: Multi-graph, Hierarchical graph, Time series graph, and Sparse graph.
• Review four multi-graph variants (Horizontal&Vertical, Temporal&Frequency, Local&Global) and two hierarchical variants (Dynamic and Predetermined).
• Discuss future directions including temporal full-connected graphs, graph condensation, heterogeneous graphs, and dynamic graphs.

Experimental results

Research questions

• RQ1 How can GNNs for EEG-based emotion recognition be systematically categorized within a unified graph-construction framework?
• RQ2 What are the common node features, edge constructions, and graph architectures used, and how do they relate to physiological brain patterns?
• RQ3 What are the main open challenges and promising directions for improving GNN-based EEG emotion recognition?

Key findings

• The survey provides the first comprehensive review and a unified taxonomic framework for GNNs in EEG-based emotion recognition.
• It categorizes methods by node-level features, edge computation, and graph-level structures, revealing common design choices and distinctions.
• It identifies four graph-level structures (Multi-graph, Hierarchical graph, Time series graph, Sparse graph) and multiple subtypes within each stage.
• It highlights practical future directions such as temporal full-connected graphs, graph condensation, and heterogeneous/dynamic graphs to address current limitations.