Analysis of Psychological and Emotional Tendency Based on Brain Functional Imaging and Deep Learning
When facing various pressures, human beings will have different degrees of bad psychological emotions, especially depression and anxiety. How to effectively obtain psychological data signals and use advanced intelligent technology to identify and make decisions is a research hotspot in psychology an...
Guardado en:
| Autor principal: | |
|---|---|
| Formato: | article |
| Lenguaje: | EN |
| Publicado: |
Hindawi Limited
2021
|
| Materias: | |
| Acceso en línea: | https://doaj.org/article/8ec919e547e3437fb1a055ded12b290f |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: | When facing various pressures, human beings will have different degrees of bad psychological emotions, especially depression and anxiety. How to effectively obtain psychological data signals and use advanced intelligent technology to identify and make decisions is a research hotspot in psychology and computer science. Therefore, a personal emotional tendency analysis method based on brain functional imaging and deep learning is proposed. Firstly, the EEG forward model is established according to functional magnetic resonance imaging (fMRI), and the transfer matrix from the signal source at the cerebral cortex to the head surface electrode is obtained. Therefore, the activation results of fMRI emotional experiment can be mapped to the three-layer head model to obtain the EEG topographic map reflecting the degree of emotional correlation. Then, combining data enhancement (Mixup) with three-dimensional convolutional neural network (3D-CNN), an emotion-related EEG topographic map classification method based on M-3DCNN is proposed. Mixup is used to generate virtual data, the original data and virtual data are used to train the network together, the number of training samples is expanded, the overfitting phenomenon of 3D-CNN is alleviated, and 3D-CNN is used for feature extraction and classification. Experimental data analysis shows that, compared with traditional methods, the proposed method can retain emotion related EEG signals to a greater extent and obtain a higher accuracy of emotion five classifications under the same feature dimension. |
|---|