Robust Support Matrix Machine for Single Trial EEG Classification.

Area of Science:

• Neuroscience
• Machine Learning
• Signal Processing

Background:

• Electroencephalogram (EEG) signals are complex and often represented as matrices.
• EEG signal classification is crucial but challenging due to artifacts and noise.
• Existing matrix classifiers struggle with contaminated EEG data.

Purpose of the Study:

• To propose a novel robust support matrix machine (RSMM) for classifying single-trial EEG data in matrix form.
• To address the challenge of intra-sample outliers and noise in EEG signals.
• To develop a classifier that accounts for the inherent low-rank structure of clean EEG data.

Main Methods:

• Proposed a robust support matrix machine (RSMM) classifier for matrix-form EEG data.
• Assumed EEG matrices decompose into a low-rank clean component and a sparse noise component.
• Developed a unified framework with an alternating direction method of multipliers (ADMM) solver for simultaneous signal recovery and classification.

Main Results:

• Extensive experiments on real binary EEG signals demonstrated superior performance.
• The RSMM method outperformed existing state-of-the-art matrix classifiers.
• The approach effectively handles intra-sample outliers and noise in EEG signals.

Conclusions:

• The proposed RSMM offers a robust solution for EEG signal classification in the presence of noise and outliers.
• This method can enhance the development of reliable brain-computer interfaces (BCIs).
• Promotes the broader adoption of noninvasive BCI technologies through improved motor imagery classification.