An Efficient Automatic Gait Anomaly Detection Method Based on Semisupervised Clustering

Algorithm 2

The pseudocodes of the BC-COP-K-means algorithm.

	Algorithm: BC-COP-K-Means
	Input: (1) Disjoint λ sets of data points with preassigned class labels
	(2) Set of all data points (except the data point of the new individual) ,
	(3) The data point of the new individual xt,
	(4) Number of clusters K.
	Output: the class label of xt
	1. Calculate the upper boundary vector up, and lower boundary vector lp of the feature vectors of each Qp
	2.  If xt falls into the area surrounded by the up and lp of a certain class Qp,
	2.1. Xt = Xt − 1 ∪ {xt}
	2.2. Return class label p
	else
	2.3. Go to step 3
	3.   Generate the must-link constraints ML and cannot-link constraints CL with .
	4.   Merge the dataset
	5.   Use the COP-K-means algorithm to generate disjoint K partitioning of the dataset Xt′
	6.   Return class label xt