Pruning Training Samples Using a Supervised Clustering Algorithm

As practical pattern classification tasks are often very-large scale and serious imbalance such as patent classification, using traditional pattern classification techniques in a plain way to deal with these tasks has shown inefficient and ineffective. In this paper, a supervised clustering algorithm based on min-max modular network with Gaussian-zero-crossing function is adopted to prune training samples in order to reduce training time and improve generalization accuracy. The effectiveness of the proposed training sample pruning method is verified on a group of real patent classification tasks by using support vector machines and nearest neighbor algorithm.