Transfer Kernel Common Spatial Patterns for Motor Imagery Brain-Computer Interface Classification

Algorithm 2

TKL algorithm.

Input: Data ; kernel ; eigen-damping factor .

Output: Domain-invariant kernel .

(1) Compute the target kernel function , the source kernel

function , and cross-domain kernel function

by kernel .

(2) Eigendecompose the target kernel function for the

eigensystem by (5).

(3) Extrapolate for source eigenvector via Nystrm

approximation by (6).

(4) Generate the extrapolated source kernel by (7)

(5) Minimize the approximation error between the ground

truth source kernel and the extrapolated source

kernel by (8).

(6) Solve QP problem (9) for eigenspectrum

(7) Construct domain-invariant kernel by (11).