• DocumentCode
    2520618
  • Title

    EBP-learning algorithm for multi-layered and inter-connected neural networks

  • Author

    Yamamoto, Yoshihiro

  • Author_Institution
    Fac. of Eng., Tottori Univ., Japan
  • fYear
    1998
  • fDate
    29-31 Jul 1998
  • Firstpage
    803
  • Lastpage
    808
  • Abstract
    The EBP algorithm has been proposed by the author for a multi-layered neural network without using a gradient method. This algorithm consists of two steps. First, fictitious teacher signals for the outputs of each hidden layer unit are algebraically determined by an error backpropagation (EBP) method. Then, the weight parameters are determined by using an orthogonal projection (EBP-OP) method, or an exponentially weighted least squares (EBP-EWLS) method. It is shown that the algorithm is also applicable for an inter-connected neural network
  • Keywords
    backpropagation; multilayer perceptrons; error backpropagation; exponentially weighted least squares method; fictitious teacher signals; inter-connected neural networks; multi-layered neural networks; orthogonal projection; Computer networks; Control systems; Electronic mail; Gradient methods; Knowledge engineering; Least squares methods; Multi-layer neural network; Neural networks; Pattern recognition; Supervised learning;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    SICE '98. Proceedings of the 37th SICE Annual Conference. International Session Papers
  • Conference_Location
    Chiba
  • Type

    conf

  • DOI
    10.1109/SICE.1998.742918
  • Filename
    742918