Multiple inputs using NEWFF

Question

Joe 2011-3-16

1
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/3325-multiple-inputs-using-newff

I am writing a program to generate a neural network using NEWFF.

I have 5 input variables (A through E) which are currently being input as an array:

input =  [ A1 A2 A3 A4 A5
           B1 B2 B3 B4 B5
           C1 C2 C3 C4 C5
           D1 D2 D3 D4 D5
           E1 E2 E3 E4 E5 ];

There is 1 target vector:

target = [ T1 T2 T3 T4 T5 ];

I input this to NEWFF as:

net = newff( input, target );

I would think this would generate a network with 5 inputs and 1 output. However, no matter how I format the input variables the network is defined as having only 1 input.

Can anyone clarify what the issue is?

Thanks, Joe

P.S. I am using R2009B

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Catherine DA GRACA 2011-4-10

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/3325-multiple-inputs-using-newff#answer_7192

在 MATLAB Online 中打开

Your input array only has one row in it - MATLAB ignores the line breaks. Use semi-colons to separate the values:

input = [ A1 A2 A3 A4 A5;
         B1 B2 B3 B4 B5;
         C1 C2 C3 C4 C5;
         D1 D2 D3 D4 D5;
         E1 E2 E3 E4 E5 ];

2 个评论
显示无隐藏无

Walter Roberson 2011-4-10

Not correct, Catherine.

>> [1 2 3 4 5

6 7 8 9 10

11 12 13 14 15]

ans =

1 2 3 4 5

6 7 8 9 10

11 12 13 14 15

When a line break is hint in an array and there is no continuation marker, it is treated as a vertical concatenation.

Joe 2011-4-11

Catherine, the intent was not to convey the exact way I'd written the code, it was more to convey the fact that it I would have thought a 5 by 5 matrix would result in a system of 5 inputs, each consisting of elements, where it appears that that is not the case. In fact, the way the matrix is built in the code is much more complex.

请先登录，再进行评论。

Answer 2

Walter Roberson 2011-4-10

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/3325-multiple-inputs-using-newff#answer_7197

Please recheck the documentation:

T SN x Q2 matrix of Q2 sample SN-element target vectors

Your target is 1 x 5 so SN is 1, so you are defining a network with a single input.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Joe 2011-4-12

Thank you for your response Walter, but could you please elaborate? The way I read that line of the documentation is that The number of rows (SN) determines the number of elements in each output, while the number of columns (Q2) determines the number of samples per output... This does not impact the number of inputs fed to the network, and a quick test of the function bears this out (a dummy set of matrices run through newff, one with a T matrix which is 1x3 and the other which is 3x1, both result in a network with only 1 input)

Considering the variable P from the documentation, which is where I would expect the number of inputs to be determined:

P: R x Q1 matrix of Q1 sample R-element input vectors

R is the number of elements in each input and Q1 is number of samples... and no matter what the structure of this matrix is it doesn't seem to impact the "number of inputs".

I think I'm missing something fundamental here.

请先登录，再进行评论。