全文下载： 
202104016.pdf
文章编号： 1672-6987（2021）04-0113-06； DOI： 10.16351/j.1672-6987.2021.04.016
张俊虎， 刘赟玥， 王玲玲， 袁栋梁
(青岛科技大学 信息科学技术学院，山东 青岛 266061）
摘要： 针对传统BP神经网络存在的容易产生过拟合、网络计算耗时长等问题，提出基于属性核特征选择与隐含层节点数动态确定的BP神经网络模型（AC-H-BPNN）。该模型以粗糙集中属性核为基点，通过属性重要度的判断，对神经网络输入项进行降维约简。针对隐含层节点数难以确定的问题，将二分分割法与经验公式相结合，精准确定隐含层节点数。并以水产养殖中对虾产量为例进行分析，实验结果表明，改进后的算法能够克服局部最小值问题，且预测结果准确度较高。
关键词： 属性核； 二分分割； BP神经网络； 隐含层节点数; 水产养殖
中图分类号： TP 183文献标志码： A
引用格式： 张俊虎， 刘赟玥， 王玲玲， 等. 基于属性核特征选择与隐含层节点数动态确定的BP神经网络模型［J］. 青岛科技大学学报(自然科学版), 2021, 42(4）: 113-118.
ZHANG Junhu，LIU Yunyue，WANG Lingling，et al. BP neural network model based on attribute core feature selection and dynamic determination of number of hidden layer nodes［J］. Journal of Qingdao University of Science and Technology（Natural Science Edition), 2021， 42（4）： 113-118.
BP Neural Network Model Based on Attribute Core Feature Selection and
Dynamic Determination of Number of Hidden Layer Nodes
ZHANG Junhu， LIU Yunyue， WANG Lingling， YUAN Dongliang
(College of Information Science and Technology, Qingdao University of Science and Technology, Qingdao 266061， China)
Abstract: Aiming at the problems of traditional BP neural network, such as overfitting and time-consuming network calculation. This paper proposes a BP neural network model based on attribute kernel feature selection and dynamic determination of the number of hidden layer nodes(AC-H-BPNN). This paper takes the attribute kernel of rough set as the model base point, and reduces the dimensionality reduction of the neural network input items through the judgment of attribute importance. Aiming at the problem that the number of hidden layer nodes is difficult to determine, the dichotomy method is combined with an empirical formula to accurately determine the number of hidden layer nodes. Taking shrimp production in aquaculture as an example for analysis, the experimental results show that the improved algorithm can overcome the local minimum problem, and the prediction result has high accuracy.
Key words: attribute core； dichotomy； BP neural network； hidden layer hodes; aquaculture
收稿日期： 2020-07-15
基金项目： 国家自然科学基金青年基金项目(61802217).
作者简介： 张俊虎(1974—),男,副教授.
