Abstract: Although it is rapid and efficient to use the grid-based clustering approach to learn the partition of a data set,grid clustering is excessively dependent on the initialization of density threshold, and the margin of each cluster constructed by the approach presents zigzag manner, which prohibits the recognition of smooth boundary surface. Thus, this paper proposed a new grid-oriented cluster ensemble approach(RG) to solve this problem. Instead of constructing the partitions with diversity on a given data set by random sampling or initializing parameters of corresponding algorithm,RG randomly splits the features set into K subsets,uses feature transformation method on the subsets to learn K differrent rotation basis,and applies grid cluster algorithm to the new feature space formed by the K axis rotations to learn the partitions with diversities. Experimental results show that, compared with single grid clustering, RG not only partilions the data set with arbitrary shape or size efficiently, but also alleviates its dependence on the density threshold initialization and smoothes the rough boundary.

Key words: Grid clustcring,Clustering algorithm,Clustcring ensemble