Weighted Learning for Feedforward Neural Networks

在线阅读 下载PDF 导出详情
摘要 Inthispaper,weproposetwoweightedlearningmethodsfortheconstructionofsinglehiddenlayerfeedforwardneuralnetworks.Bothmethodsincorporateweightedleastsquares.Ourideaistoallowthetraininginstancesnearertothequerytoofferbiggercontributionstotheestimatedoutput.Byminimizingtheweightedmeansquareerrorfunction,optimalnetworkscanbeobtained.Theresultsofanumberofexperimentsdemonstratetheeffectivenessofourproposedmethods.
机构地区 不详
出版日期 2014年03月13日(中国期刊网平台首次上网日期,不代表论文的发表时间)
  • 相关文献