大数据挖掘中神经网络学习算法高可靠性仿真

High-Reliability Simulation of Neural Network Learning Algorithm In Big Data Mining

大数据挖掘是从大量的随机数据中,挖掘出潜藏在数据库中有用的知识的过程。在挖掘的过程中,若迭代次数过多,会导致其并行效率降低,严重影响了挖掘的性能。为了有效解决上述问题,提出一种大数据挖掘中神经网络学习算法的可靠性方法。采用正交基函数处理网络输入函数和连接权函数,将结构参数和其它参数整合为一个粒子,使用粒子优化算法(PSO)对全局优化处理。通过组建大数据挖掘的神经网络训练模型,展开数据特征提取和聚类处理,全面提升大数据挖掘过程中的聚类以及特征提取能力。实验结果表明,所提方法的挖掘准确率在94%以上,具有高可靠性的数据挖掘结果。

Big data mining is a process of mining useful knowledge hidden in databases from massive random data.In this process,too many iterations may lead to the reduction of parallel efficiency,thus seriously affecting the performance of mining.In order to effectively solve this problem,this paper put forward a neural network learning algorithm with high reliability based on big data mining.At first,the orthogonal basis function was used to process the network input function and connection weight function.Then,the structural parameters and other parameters were integrated into a particle.Meanwhile,the particle optimization algorithm(PSO)was used for global optimization.After building a neural network training model for big data mining,we carried out the data feature extraction as well as the clustering in order to comprehensively improve the ability of clustering and feature extraction in the process of big data mining.Experimental results show that the mining accuracy of the proposed method is more than 94%,and the data mining results are highly reliable.