Cross Domain Correlation Maximization for Enhancing the Target Recognition of SSVEP-Based Brain-Computer Interfaces.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Steady-state visual evoked potential (SSVEP)-based brain-computer interfaces (BCIs) show improved performance with training.
• Training is time-consuming and fatiguing, necessitating limited data.
• Limited training data can reduce classification performance in SSVEP BCIs.

Purpose of the Study:

• To propose a novel method to improve SSVEP BCI classification accuracy without increasing training time.
• To address the challenge of reduced classification performance due to limited training data.

Main Methods:

• Proposed a transfer-related component analysis (TransRCA) method.
• Extracted SSVEP-related components from limited individual training data and combined them with extensive existing data.
• Maximized inter-trial covariances and correlations between reference and SSVEP signals.

Main Results:

• Validated TransRCA on SSVEP Benchmark and BETA datasets.
• The ensemble version of TransRCA demonstrated superior classification accuracy and information transmission rate compared to state-of-the-art methods.
• Outperformed eCCA, eTRCA, ttCCA, LSTeTRCA, and eIISMC.

Conclusions:

• TransRCA significantly enhances SSVEP BCI performance with minimal training data.
• The method shows high potential for developing efficient SSVEP-based BCIs.
• Offers a solution for high-performance BCIs with reduced training burden.