A hierarchical clustering algorithm with multiple target shapes

In this paper, the authors present a new approach in which the multiple target shapes (MTS) for a cluster are created during the hierarchical clustering. Why MTS is needed is explained and how to generate MTS is described. They have analyzed the complexity of the algorithm and proposed a simplified method because of the heavy complexity. The results are very encouraging when it is used in BBL (building block layout) placement of the PANDA system in comparison with BEAR, developed at U.C. Berkeley. >