基于部分连接神经网络的序列数据分类算法研究

Research on sequence data classification algorithm based on partially connected neural network

研究梳理了序列数据的定义,并且分析了5种类型的序列数据,结合局部连接神经网络的特点,研究了基于局部连接神经网络的序列数据的分类算法。使用该算法进行计算,学习与收敛速度较快,对于自适应建模与控制十分适用,利用方形基函数进行计算,在网络输出过程中注意只能利用方形函数来逼近光滑函数。由于序列数据分类运算在数据挖掘中存在巨大的优势,因此对序列数据算法的研究具有很高的理论与应用价值。

The definition of sequence data is studied. The 5 kinds sequence data is analyzed. In combination with the characteristics of the partially connected neural network,the classification algorithm of the sequence data based on partially connected neural network is studied. The algorithm used to calculation has fast learning and convergence rate,and is especially suitable for adaptive modeling and control. When the square primary function is used to calculate,the square function can be only used to approach the smooth function in network output process. The sequence data classification operation has great advantage in data mining,so the research of the sequence data classification algorithm has the high theory and application value.