计算机技术与发展2017,Vol.27Issue(2):155-157,162,4.DOI:10.3969/j.issn.1673-629X.2017.02.035
梯度神经网络在求解矩阵平方根中的应用
Application of Gradient Neural Network in Matrix Square Root Solving
摘要
Abstract
Matrix square root problem can be regarded as a special case of matrix problems,and has a wide application in scientific and engineering fields.Different from the conventional numerical methods,the gradient neural network is adopted to solve matrix square root problem.In order to solve the square root of a matrix,a norm-based scalar-valued energy function is defined.Then,according to the graclient descent method,an evolution formula is designed.Thus,the gradient neural network is derived for finding the square root of a matrix by expanding the evolution formula.With the aid of computer simulation based on MATLAB,the simulation results confirm the accuracy and validity of the gradient neural network for finding matrix square root.Furthermore,by choosing different values of the design parameter,the convergence streed of the gradient neural network for matrix square root solving has been improved greatly.The results show that design parameter plays an important role in the gradient neural network for solving matrix square root.关键词
梯度神经网络/梯度下降法/矩阵平方根/MATLAB仿真Key words
gradient neural network/gradient descent method/matrix square root/MATLAB simulation分类
信息技术与安全科学引用本文复制引用
严慧玲,肖林,周文辉..梯度神经网络在求解矩阵平方根中的应用[J].计算机技术与发展,2017,27(2):155-157,162,4.基金项目
国家自然科学基金资助项目(61503152) (61503152)
湖南省自然科学基金(2016JJ2101) (2016JJ2101)
湖南省教育厅优秀青年项目(15B192) (15B192)
吉首大学2015年实验教学改革研究项目(2015SYJG034) (2015SYJG034)
吉首大学2016年研究生科研创新项目(JGY201643) (JGY201643)
吉首大学2016年校级课题(Jdy2016009) (Jdy2016009)
吉首大学2016年大学生研究性学习和创新性实验计划项目资助(教通[2016]13号) (教通[2016]13号)