A New Clustering Algorithm Using Attribute Boosting
Cheng-An Li · 2009
Combining multiple clusterings is an effective technique for improving clustering accuracy. In this paper, a new boosting clustering algorithm is proposed to improve the performance of clustering of data sets. In this algorithm, attribute subsets are extracted based on their importance. The method alters the distribution by emphasizing particular attributes while standard boosting algorithms alter the distribution by emphasizing particular training examples. A distribution over the attributes is updated at each iteration by conducting a sensitivity analysis and the attributes used by the model learned in the current iteration. The experimental results on several real-life and benchmark data sets show that the proposed algorithm can effectively improve the accuracy of clustering.