首页|引入神经网络极限学习机的关键数据查询模型

引入神经网络极限学习机的关键数据查询模型

扫码查看
网络空间数据的结构具有较高相似性,海量数据的不断增量更新,导致关键数据查询结果存在冗余和偏离问题.因此提出基于神经网络极限学习机的关键数据查询方法.建模描述关键数据查询问题.基于此引入神经网络极限学习机,建立关键数据查询模型.预处理数据库中无用数据和重复数据做,通过输出权值范数的最小二乘解,避免算法陷入局部最优.结合输出矩阵,训练查询模型,输出结果结果即为关键数据查询结果.为证明上述方法的性能优势,设计对比实验,结果表明提出的方法应用于关键数据查询的均方根误差不超过1.2,平均绝对百分比误差最高为4.1%,关系数F可达0.6,网络节点的使用率低于20%.以上实验数据验证了上述方法数据查询精度较高,可应用性更强.
A Key Data Query Model for Introducing Neural Network Extreme Learning Machine
The structure of cyberspace data has high similarity,and the continuous incremental updates of massive data lead to redundancy and deviation issues in key data query results.Therefore,a method of querying key data based on neural network and extreme learning machine was put forward.Firstly,the problems of key data query were described by modeling.On this basis,the neural network and extreme learning machine were used to construct a model of key data query.Secondly,useless data and duplicate data in database were preprocessed.And then,the least square solution of weight norm was output to prevent the algorithm from being got in local optimization.Combined with the output matrix,the query model was trained.Finally,the output result was the result of key data query.In order to prove the performance advantage of the proposed method,a comparative experiment was designed.Experimental results show that the root mean square error of key data query is less than 1.2 after using the proposed method,and the maximal mean absolute percentage error is 4.1%.In addition to these data,the relationship number F can reach 0.6.And the utilization rate of network nodes is always less than 20%.The experimental data above prove that the proposed method has higher accuracy of data query and stronger applicability.

Neural network limit learning machineKey dataOutput weightLeast square solutionData prepro-cessing

张勇飞、陈艳君、赵世忠

展开 >

南昌大学科学技术学院,江西 九江 332020

南昌大学工程建设学院,江西 南昌 330031

神经网络极限学习机 关键数据 输出权值 最小二乘解 数据预处理

江西省教育厅科学技术研究项目(2021)

GJJ217813

2024

计算机仿真
中国航天科工集团公司第十七研究所

计算机仿真

CSTPCD
影响因子:0.518
ISSN:1006-9348
年,卷(期):2024.41(3)
  • 19