Error perception classification in Brain-Computer Interfaces using CNN.

Area of Science:

• Neuroscience
• Computer Science
• Biomedical Engineering

Background:

• Accurate error perception in human-Brain-Computer Interface (BCI) interaction is crucial for system improvement and seamless user experience.
• Convolutional Neural Networks (CNNs) have shown promise in BCI error detection, eliminating the need for manual feature selection.

Purpose of the Study:

• To propose a novel CNN model for capturing human error perception in BCI systems.
• To develop a model with shorter temporal input for enhanced real-time BCI application usability.
• To evaluate and compare the proposed model against existing CNN approaches.

Main Methods:

• Development of a new CNN model with reduced temporal input.
• Evaluation using the Monitoring Error-Related Potential dataset.
• Comparative analysis against other recent CNN models.

Main Results:

• The proposed CNN model achieved an accuracy of 80%.
• Sensitivity and specificity were recorded at 76% and 85%, respectively.
• The model demonstrated superior performance compared to previous BCI error perception models.

Conclusions:

• The novel CNN model effectively captures human error perception in BCI.
• The model's design enhances usability for real-time BCI applications.
• The findings suggest a significant advancement in BCI error detection technology.