Introducing chaos behavior to kernel relevance vector machine (RVM) for four-class EEG classification.

Area of Science:

• Neuroscience
• Machine Learning
• Signal Processing

Background:

• Electroencephalography (EEG) signal classification is crucial for Brain-Computer Interfaces (BCI).
• Human brain signals exhibit complex, chaotic characteristics.
• Relevance Vector Machines (RVM) are effective for classification tasks.

Purpose of the Study:

• To develop and evaluate a novel chaos kernel function for RVM in EEG signal classification.
• To leverage the chaotic dynamics of brain signals to improve BCI performance.
• To enhance the classification capacity of RVM for motor imagery tasks.

Main Methods:

• A novel kernel function was derived from a chaotic system.
• The chaos kernel was integrated into a Relevance Vector Machine (RVM).
• The method was validated using a one-versus-one Common Spatial Pattern (OVO-CSP) classifier on a public EEG dataset for motor imagery classification.

Main Results:

• The proposed chaos kernel function demonstrated higher classification accuracy compared to Gaussian and Polynomial kernels.
• The integration of chaotic dynamics significantly improved RVM performance in EEG classification.
• The OVO-CSP framework effectively classified four types of motor imagery movements.

Conclusions:

• The developed chaos kernel function is effective for EEG signal classification in BCI applications.
• Incorporating chaotic dynamics into kernel functions enhances machine learning model performance for brain signal analysis.
• This approach offers a promising direction for advancing BCI technology through improved signal processing and classification.