Improved Transductive Support Vector Machine for a Small Labelled Set in Motor Imagery-Based Brain-Computer Interface

Algorithm 2

The proposed IST-TSVM algorithm.

	Input: the initial labelled set: .
	Input: the remaining unlabelled set: .
	Output: the labels assigned to the unlabelled samples: .
	Initialize the expanded labelled set: ;
	for  = 1 to 5 do
	Compute the CSP matrix with the expanded labelled set ;
	Calculate the CSP features of all samples using ;
	Generate the comprehensive features of all samples, as described in Section 2.2.1;
	Perform the ITSVM classifier given in Algorithm 1 to obtain the optimal parameter group using the comprehensive features of all samples;
	Calculate temporal labels of all samples based on and ;
	if then
	or
	then
	break;
	end if
	end if
	Update the labels of the unlabelled samples
	Construct the selected unlabelled set based on the confidence criterion in Section 2.3.1
	Expand the labelled set: ;
	end for