Related Experiment Video
Updated: Mar 25, 2026
05:48
Author Spotlight: Investigating the Impact of Emotional Prosodies on Voice Recognition and Perception
Published on: August 9, 2024
2.1K
CBR-Net: A Multisensory Emotional Electroencephalography (EEG)-Based Personal Identification Model with
Rui Ouyang1, Minchao Wu2, Zhao Lv1
1Anhui Province Key Laboratory of Multimodal Cognitive Computation, School of Computer Science and Technology, Anhui University, Hefei 230601, China.
Bioengineering (Basel, Switzerland)
|March 28, 2025
Summary
Multisensory stimuli, like olfactory cues, improve electroencephalography (EEG) emotion recognition accuracy. A novel CNN-BiLSTM-Residual Network (CBR-Net) model shows superior performance in identifying individuals based on EEG signals.
Area of Science:
- Neuroscience
- Computer Science
- Biomedical Engineering
Background:
- Electroencephalography (EEG)-based personal identification is sensitive to emotional state fluctuations, impacting accuracy.
- Multisensory stimuli, including video and olfactory cues, can modulate emotional responses and potentially enhance EEG-based identification.
Purpose of the Study:
- To propose a novel deep learning model, CNN-BiLSTM-Residual Network (CBR-Net), for improved EEG-based identification.
- To establish a multisensory emotional EEG dataset using video-only and olfactory-enhanced video stimuli.
- To evaluate the impact of olfactory enhancement on EEG signal emotional intensity and identification accuracy.
Main Methods:
- Development of the CBR-Net model integrating Convolutional Neural Network (CNN) for spatial features, Bi-LSTM for temporal dynamics, and residual connections.
- Creation of a novel dataset comprising EEG signals recorded under video-only and olfactory-enhanced video conditions.
- Comparative analysis of CBR-Net performance against traditional machine learning and other deep learning models.
Main Results:
- Olfactory-enhanced video stimulation significantly increased EEG signal emotional intensity, leading to improved recognition accuracy.
- CBR-Net achieved high accuracy across emotions: 96.59% for negative, 95.42% for positive, and 94.25% for neutral.
- Ablation studies indicated Bi-LSTM's importance for neutral emotions and CNN's for positive emotions.
- CBR-Net outperformed existing models in EEG-based identification across all emotional states.
Conclusions:
- CBR-Net effectively enhances identity recognition accuracy using EEG signals.
- Multisensory stimuli, particularly olfactory enhancement, offer significant advantages for improving EEG-based identification.
- The findings support the integration of multisensory emotional stimulation in biometric systems.
