集成学习中基于离散化方法的基分类器构造研究

为构造集成学习中具有差异性的基分类器，提出基于数据离散化的基分类器构造方法，并用于支持向量机集成。该方法采用粗糙集和布尔推理离散化算法处理训练样本集，能有效删除不相关和冗余的属性，提高基分类器的准确性和差异性。实验结果表明，所提方法能取得比传统集成学习算法Bagging和Adaboost更好的性能。

Research on construction of base classifiers based on discretization method for ensemble learning

Construction method of base classifiers based on data discretizaion was proposed to produce individual classifiers with good diversity in ensemble learning. And then it was used in support vector machines ensemble. Using the rough sets and Boolean reasoning algorithm to process the training samples, this method can eliminate the irrelative and redundant attributes to improve the accuracy and diversity of base classifiers. Experimental results show that the presented method can achieve better performance than the traditional ensemble learning methods such as Bagging and Adaboost.