深度神经网络和Bagging分类方法的随机模拟与实证研究

Stochastic Simulation and Empirical Research of Deep Neural Networks and Bagging Classification Methods

摘要: 本文利用随机模拟实验和实证研究，考虑了深度神经网络分类和Bagging分类两种机器学习方法的建模和预测。首先，本文通过对比两种模型定义、相关的联系和区别，给出了两种方法各自的适用场景及优势和不足。之后随机生成非线性数据集进行随机模拟，结果表明，神经网络分类模型具有较高的精确度。进一步，本文选取较优的神经网络分类模型对一个实际数据集：鸢尾花分类数据集进行实证研究，并进行相关预测，结果表明，神经网络分类模型具有较高的准确度与预测精度，对于分类器的选择问题具有重要的现实意义。

Abstract: This paper uses stochastic simulation experiments and empirical research, and considers the mod-eling and prediction of two machine learning methods: deep neural network classification and Bag-ging classification. First, by comparing the definitions, related connections and differences of the two models, this paper presents the applicable scenarios and advantages and disadvantages of the two methods. Then the nonlinear dataset was randomly generated for stochastic simulation, which showed that the neural network classification model has high accuracy. Further, this paper selects an excellent neural network classification model to conduct an empirical research on a real data set: Iris classification data set, and makes relevant predictions. The results show that the neural net-work classification model has high accuracy and prediction accuracy, which has important practical significance for the problem of classifier selection.