Support vector machine based hierarchical clustering of spatial databases

Support vector machine (SVM) is applied to recognize two separable classes. The algorithm builds up a binary tree to tackle multi-class recognition by SVM based hierarchical clustering. SVM is used to recognize two classes and builds up a binary tree in a bottom-to-up version to analyze the multi-class recognition. The number of binary trees is ultimately the number of clustering. It can be applied to clustering problems of arbitrary shapes, achieving the best result, and adapted to fit for the analysis of high dimensional data.