ITopDownClustering.java

package clustering;

import java.util.List;

public interface ITopDownClustering<ClusterType, ClusterMemberType> {

	/**
	 * Can we split a cluster into two.
	 * @param c
	 * @return
	 */
	public boolean canSplitFrom(ClusterType c1, ClusterType c2,
		ISimilarity<ClusterMemberType> simFunc, double simThreshold);

	/**
	 * Supports k-way splits. This can be achieved via graph cut algorithms, 
	 * or the simplest method being k-means. Question is how to compute the 
	 * median in k-means over activity frames. 
	 * @param c
	 * @return
	 */
	abstract public List<ClusterType> splitACluster(ClusterType c, int k) throws Exception;

	public static interface IStopping<ClusterType> {

		public boolean split(ClusterType parent, List<ClusterType> children);
	}
}