A Least-Square Unified Framework for Spatial Filtering in SSVEP-Based BCIs.

Area of Science:

• Biomedical Engineering
• Machine Learning
• Neuroscience

Background:

• Steady-state visual evoked potential (SSVEP) is a key Brain-Computer Interface (BCI) paradigm known for its high information transfer rate.
• SSVEP finds extensive use in assistive and rehabilitation technologies.

Purpose of the Study:

• To propose a unified least-square (LS) framework for analyzing correlation analysis (CA)-based SSVEP spatial filtering methods.
• To offer a machine learning perspective on these methods, clarifying their commonalities, differences, and computational factors.

Main Methods:

• Developed a generalized optimization problem to determine spatial filters, incorporating non-linear and regularization terms.
• Performed a comparative analysis of existing SSVEP spatial filtering techniques.
• Integrated recommended design strategies to address research gaps and foster algorithmic advancements.

Main Results:

• The LS framework provides an intuitive interpretation of computational factors in SSVEP spatial filtering.
• Identified superior and robust design strategies for spatial filtering.
• Developed five novel spatial filtering methods based on the LS framework and integrated strategies.

Conclusions:

• The proposed LS framework offers significant insights into the relationships between spatial filtering design strategies from a machine learning viewpoint.
• This work contributes to the advancement of high-performance SSVEP recognition methods for BCI applications.