Related Experiment Video
Updated: May 14, 2025

09:44
Author Spotlight: Advancing Large-Scale Neural Dynamics Through HD-MEA Technology
Published on: March 8, 2024
4.5K
MSDSANet: Multimodal Emotion Recognition Based on Multi-Stream Network and Dual-Scale Attention Network Feature
Weitong Sun1,2,3, Xingya Yan1,2, Yuping Su3,4
1School of Digital Art, Xi'an University of Posts & Telecommunications, Xi'an 710061, China.
Sensors (Basel, Switzerland)
|April 12, 2025
Summary
This study introduces a novel multimodal emotion recognition model using electroencephalography (EEG) and electrooculography (EOG) signals. The advanced model enhances feature representation and spatiotemporal modeling for more accurate emotion recognition.
Area of Science:
- Neuroscience
- Computer Science
- Artificial Intelligence
Background:
- Current electroencephalography (EEG) emotion recognition models face limitations in feature representation granularity and spatiotemporal dependence modeling.
- Effective emotion recognition requires capturing complex patterns in brain and eye activity.
Purpose of the Study:
- To propose a novel multimodal emotion recognition model that addresses the shortcomings of existing EEG-based approaches.
- To enhance feature representation and spatiotemporal modeling for improved emotion recognition accuracy.
Main Methods:
- A multimodal model integrating multi-scale feature representation and attention mechanisms was developed.
- The model employs a multi-stream network for shallow EEG feature extraction and a dual-scale attention module for shallow electrooculography (EOG) feature extraction.
- Multi-scale and multi-granularity feature fusion was utilized to improve feature richness and discriminability.
Main Results:
- The proposed model demonstrated superior performance compared to existing models on two benchmark datasets.
- Enhanced feature fusion led to richer and more discriminative multimodal feature representations.
- The integration of attention mechanisms improved the model's ability to capture spatiotemporal dependencies.
Conclusions:
- The developed multimodal emotion recognition model effectively overcomes limitations in EEG feature representation and spatiotemporal modeling.
- The proposed approach offers a significant advancement in the field of affective computing and emotion recognition.
- This model shows promise for real-world applications requiring accurate and robust emotion detection.

