学科分类
/ 1
1 个结果
  • 简介:Inthispaper,weproposetwoweightedlearningmethodsfortheconstructionofsinglehiddenlayerfeedforwardneuralnetworks.Bothmethodsincorporateweightedleastsquares.Ourideaistoallowthetraininginstancesnearertothequerytoofferbiggercontributionstotheestimatedoutput.Byminimizingtheweightedmeansquareerrorfunction,optimalnetworkscanbeobtained.Theresultsofanumberofexperimentsdemonstratetheeffectivenessofourproposedmethods.

  • 标签: 单隐层前馈神经网络 加权最小二乘 学习方法 误差函数 最小化 查询