neural network input target data format, Vertical or Horizontal vector within cell array?

I have a set of vectors measured dx dy on a given set of coordinates over a time period i=1:5;

If I make two neural networks, one for x and one for y, does my input for “error” and “input” (see below example code) have to be vertical or horizontal vector within the cell array?

%%%%%%%%%
example code
%%%%%%%%%

% if I have a set of vertical vectors coordinates with X position defined by Column 1 = cord(:,1) and Y position defines by Column 2 = (cord(:,2))

cord=[-2 -2;-2 -1;-2 0;-2 1;-2 2;-1 -2;-1 -1;-1 0;-1 1;-1 2;0 -2;0 -1;0 0;0 1;0 2; 1 -2;1 -1;1 0;1 1;1 2;2 -2;2 -1;2 0;2 1;2 2];

% I have a set of target vectors per point defined by dx = tmpError(:,1) & dy = tmpError(:,2)

tmpError=[0.3 0.2;0.1 0.2;0.3 0.4;0.1 0.2;0.4 0.5;...
   0.1 0.4;0.2 0.1;0.3 0.2;0.1 0.2;0.4 0.5;...
   0.2 0.2;0.4 0.4;0.2 0.1;0.3 0.2;0.1 0.5;...
   0.4 0.3;0.3 0.3;0.3 0.2;0.2 0.2;0.3 0.5;...
   0.3 0.2;0.4 0.2;0.1 0.4;0.4 0.2;0.4 0.5];

for i=1:5
  error{i}=tmpError*i;
  errorDx{i}=tmpError*i;
  errorDy{i}=tmpError*i;    
end

% Input would be same dxInput = input(:,1) & dyInput = input(:,2)

tmpInput=[1.3 1.2;1.1 1.2;1.3 1.4;1.1 1.2;1.4 1.5; 2.1 2.4;2.2 2.1;2.3 2.2;2.1 2.2;2.4 2.5; 3.2 3.2;3.4 3.4;3.2 3.1;3.3 3.2;3.1 3.5; 4.4 4.3;4.3 4.3;4.3 4.2;4.2 4.2;4.3 4.5; 5.3 5.2;5.4 5.2;5.1 5.4;5.4 5.2;5.4 5.5];

for i=1:5
  input{i}=tmpInput*i;
  inputDx{i}=tmpInput(:,1)*i;
  inputDy{i}=tmpInput(:,2)*i;
end

%% OR should they be transposed....

cord=cord';
tmpError=tmpError';
for i=1:5
    error{i}=tmpError*i;
    errorDx{i}=tmpError(1,:)*i;
    errorDy{i}=tmpError(2,:)*i;
end

tmpInputDx=repmat(tmpInput(:,1),1,5);
tmpInputDy=repmat(tmpInput(:,2),1,5);

[Greg Heath]

AAARRRGH ... Please use the ANSWERS formatting rules. See the blocks B,I,Aa,...{} Code and Help above the reply box? Even if you just use 1 indented line per command, it would help immensely. Also try putting braces {} around your code.

Greg

[Emil]

Thanks for your reply.

I fixed the original code (i hope) ... See code in original message.

Answer 1

[DemoiselX]

hi, can you explain more your problem ?

[Emil]

Sure.. sorry for not coming across clear enough.

If I use cell array as my input and target a Neural Net Time Series, I understand that the input & target can be single column (within a row of a cell array) per point in time or a single row with multiple columns. If all inputs effect traget postions equally I would think it would be ok to enter them as a single column. It not the same could be done by transposing my original vector an repmat

as well, just realized my example code should looks something like this for the transposed input....

tmpInputDx=repmat(tmpInput(:,1),1,5)

Answer 2

[Greg Heath]

It would be helpful to the reader (especially the older ones) if you would use the notation 0.69 instead of just .69.

All inputs and outputs to the NN functions are matrices or cells of column vectors.

For a data set of N I-dimensional input vectors and N corresponding O-dimensional target vectors

 [ I N ] = size(input)

 [ O N ] = size(target)

or the cell equivalents

 Inputs  = { inputs };

 Targets = { targets };

Hope this helps.

Thank you for formally accepting my answer

Greg

[Greg Heath]

What is the purpose of multiplying by i?

When you post code make sure that it will not throw errors when cut and pasted into the command line. For example,

for i=1:5 input{i}=tmpInput*i; end

        |
Error: The expression to the left of the equals sign is not
a valid target for an assignment.

You omitted the comma or semicolon after the 5.

Also

cord=[-2 -2;-2 -1;-2 0;-2 1;-2 2;-1 -2;-1 -1;-1 0;-1 1;-1 2;0 -2;0 -1;0 0;0 1;0 2; 1 -2;1 -1;1 0;1 1;1 2;2 -2;2 -1;2 0;2 1;2 2];

 Error using vertcat
 CAT arguments dimensions are not consistent.

Hope this helps.

Greg

[Emil]

The fixed the example code. Sorry for the trouble.

I use the i only to for the generation of data. I could have made random numbers but took a short cut I guess.

For neural network time series you can have multiple cell array sub struct which contain vectors of data to be modeled per point in time... In my case I have a set of vectors per point in time. My question is if that set of input and targets of vectors needs to be vertical or horizontal

[Greg Heath]

In the general case, they are columns. You may be getting fooled by looking at one-dimensional timeseries. Each measurement is a one-dimensional column vector. The fact that you see the total data as a row instead of a sequence of one-dimensional columns is irrelevant.

Hope this helps.

Greg