Emotion Recognition Using Hierarchical Spatiotemporal Electroencephalogram Information from Local to Global Brain...

Area of Science:

• Neuroscience
• Machine Learning
• Biomedical Engineering

Background:

• Understanding human emotional states requires analyzing local brain activity and inter-regional interactions.
• Multichannel electroencephalography (EEG) offers a non-invasive method for capturing brain dynamics related to emotions.

Purpose of the Study:

• To propose a hierarchical emotional context feature learning model for enhanced multichannel EEG-based emotion recognition.
• To improve the learning of spatiotemporal EEG features from local to global brain regions.

Main Methods:

• A hierarchical model comprising regional and global brain-level encoding modules with a classifier.
• Inputting multichannel EEG signals grouped into nine functional brain regions.
• Utilizing two-layer and one-layer bidirectional gated recurrent units (BGRUs) with self-attention for feature extraction.

Main Results:

• The proposed model demonstrated superior performance in EEG-based emotion recognition across three datasets.
• The hierarchical approach effectively captured spatiotemporal EEG signal characteristics.
• The method outperformed existing state-of-the-art techniques.

Conclusions:

• The hierarchical emotional context feature learning model significantly improves emotion recognition from EEG signals.
• Integrating local and global brain region information is crucial for accurate emotional state identification.
• The model provides a robust framework for analyzing complex brain dynamics in emotional processing.