Formula for number of unknown weights in Recurrent Neural Network

1 次查看（过去 30 天）

Khawaja Asim 2016-1-6

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/262877-formula-for-number-of-unknown-weights-in-recurrent-neural-network

编辑： Greg Heath 2016-1-8

Given a recurrent neural network with two hidden layers, with "I" represent the number of features of inputs, "O" represent the number of outputs. H1 and H2 represent the number hidden neurons in layer 1 and 2. What will be the formula for number of synaptic connections in Recurrent Neural Networks in terms of I, O, H1 and H2.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Greg Heath 2016-1-8

编辑：Greg Heath 2016-1-8

在 MATLAB Online 中打开

I. Be specific:

   MATLAB Version: 8.3.0.532 (R2014a)
   Neural Network Toolbox  Version 8.2  (R2014a)

>> lookfor recurrent

 layrecnet - Layered recurrent neural network.
 newhop    - Design a Hopfield recurrent network.
 newlrn    - Create a Layered-Recurrent network.

II. Why don't you consider layrecnet and newlrn

 A. Read the help documentation
 B. Read the  doc documentation
 C. Modify the corresponding code examples
    1. net1 with H2=0 
    2. Consider view(net1)and   
        Nw1 = net1.numWeightElements
    3. Repeat 1 and 2 for net2 with H2~=0