基于自适应距离度量的最小距离分类器集成

提出了一种基于自适应距离度量的最小距离分类器集成方法，给出了个体分类器的生成方法。首先用Bootstrap技术对训练样本集进行可重复采样，生成若干个子样本集，应用生成的子样本集建立自适应距离度量模型，根据建立的模型对子样本集进行训练，生成个体分类器。在集成中，将结果用相对多数投票法集成最终的结论。采用UCI标准数据集实验，将该方法与已有方法进行了性能比较，结果表明基于自适应距离度量的最小距离分类器集成是最有效的。

Minimum distance classifier ensemble based on adaptive distance metric

A minimum distance classifier ensemble method based on adaptive distance metric was proposed. The training method of component classifier was given. Some training subsets were obtained via bootstrap technique, then the model about adaptive distance metric with the training subset was established. Each component classifier was trained independently using the model, then some component classifiers were obtained. After that, they were collected to make a decision according to the majority voting. Experiment results on UCI standard database show that the proposed ensemble method based on adaptive distance metric for minimum distance classifier is effective, and it is superior to other methods in classification performance.