Research Article
A Distributed Algorithm for the Cluster-Based Outlier Detection Using Unsupervised Extreme Learning Machines
Algorithm 2
CB outlier detection on each slave node.
Input. The cluster set reserved on slave node , integers , , the threshold | Output. The CB outliers in | () for each cluster in do | () Sort the points in according to the distances to the centroid point | in ascending order; | () Scan the points in reversed order; | () for each scanned point do | () Initialize a heap ; // to reserve the current NNs of | () ; // the largest distance from the points in to | () boolean = true; | () Visit the points from to the both sides to search ’s NNs; | () for each visited point do | () if is before and | then | () Erase the points before from the visited list; | () else if is behind and | then | () Erase the points behind from the visited list; | () else | () Update and ; | () if meets the condition of Corollary 5 then | () = false; | () break | () if then | () Send to the master node to update ; |
|