基于并行反向熵决策树算法的人工神经网络

NEURAL NETWORK ON PARALLEL REVERSE ENTROPY DECISION TREE

针对数据分类挖掘问题,利用并行思想,提出一种基于并行反向熵决策树算法的人工神经网络。通过概率度量水平生成并行决策树对数据进行粗处理,以加快人工神经网络的分析速度。随后采用一组仿真数据对该方法进行测试和评估。实验结果表明,该并行分类方法比单个决策树具有更高的分类精度,并在保持分类结果良好可解释性的基础上优化了分类规则。

For classification problems in data mining, based on the parallel thinking, in this paper it proposed a neural network which is based on parallel reverse entropy decision trees. To rough handle the data through level generated parallel decision trees which adopted the method of probability measurement, the analysis speed of neural network was accelerated. Then a serial of simulative data was used to test and evaluate this method. Experiment result manifested that the parallel method has higher classification accuracy level than single decision tree, and it optimizes classification rules while keeping up good interpretability for classification results.