【摘要】 朴素贝叶斯分类算法由于其计算高效在生活中应用广泛。本文根据集成算法的差异性特征,聚类算法聚类点的选择方式的可变性,提出了基于K-medoids聚类技术的贝叶斯集成算法,朴素贝叶斯的泛化性能得到了提升。首先,通过样本集训练出多个朴素贝叶斯基分类器模型;然后,为了增大基分类器之间的差异性,利用K-medoids算法对基分类器在验证集上的预测结果进行聚类;最后,从每个聚类簇中选择泛化性能最佳的基分类器进行集成学习,最终结果由简单投票法得出。将该算法应用于UCI数据集,并与其他类似算法进行比较可得,本文提出的基于K-medoids聚类的贝叶斯集成算法(NBKME)提高了数据集的分类准确率。更多还原

【Abstract】 The Naive Bayes classification algorithm is widely used in life due to its computational efficiency. Based on the difference characteristics of the ensemble algorithm and the variability of the clustering point selection method of the clustering algorithm,this paper proposes a Bayesian ensemble algorithm based on K-medoids clustering technology,and the generalization performance of Naive Bayes has been improved. Firstly,multiple Naive Bayesian classifier models are trained through the sample set; then,in order to increase the difference between the base classifiers,the K-medoids algorithm is used to gather the prediction results of the base classifiers on the validation set; Finally,the base classifier with the best generalization performance is selected from each cluster for ensemble learning,and the final result is obtained by a simple voting method. The algorithm is applied to UCI data set and compared with other similar algorithms. The Bayesian ensemble algorithm based on K-medoids clustering( NBKME)proposed in this paper improves the classification accuracy of the data set.更多还原