Graph-Enhanced Emotion Neural Decoding.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Human-Computer Interaction

Background:

• Brain signal-based emotion recognition is crucial for intelligent systems.
• Current methods often lack explicit integration of emotion-brain relationships.
• This limits the informativeness of learned representations for emotion decoding.

Purpose of the Study:

• To propose a novel graph-enhanced neural decoding approach.
• To explicitly integrate emotion-brain relationships into representation learning.
• To improve the effectiveness of emotion decoding from brain imaging data.

Main Methods:

• Developed a graph-enhanced emotion neural decoding technique.
• Utilized a bipartite graph structure to link emotions and brain regions.
• The proposed emotion-brain bipartite graph generalizes existing methods.

Main Results:

• The novel approach learns more informative representations.
• Demonstrated effectiveness and superiority on visually evoked emotion datasets.
• Theoretical analysis confirms the graph's generalization capabilities.

Conclusions:

• The proposed method enhances emotion decoding accuracy.
• Explicitly modeling emotion-brain relationships is beneficial.
• This approach advances human-computer interaction through better emotional intelligence.