sizing of the sample data in Matlab for neural network

Prev: get intensity from pixel impixelinfoval.m
Next: save Variables in Gui view

From: Rajeev Kumar on 29 Dec 2009 05:44

---

Respected Sir,
I am doing My M. tech project .I am new to matlab .I am designing feed forward backpropagation network through. I have 1 input and hundred 100 outputs. When I enter the data in matrix form e.g. Inputs=[1 2 3 4 5];
And t=[ 0 0.2 0.3………………………………………………………………………up to 100 columns;
……………………………………………………………………………………… up to 100 columns;
………………………………………………………………………………………… up to 100 columns;
……………………………………………………………………………………………up to 100 columns;
……………………………………………………………………………………………… up to 100 columns;];
Means target matrix is 5*100 i.e 5 rows and hundred columns .but the matlab gives error i.e
??? Error using ==> network.train at 228
Targets are incorrectly sized for network.
Matrix must have 100 rows.
Please help me how can size my data for input and output targets. Please reply me. My email id is er.rajeevdogra(a)rediff.com . I will be very greatful to you.
Thanking You.

From: Greg Heath on 29 Dec 2009 07:54

---

On Dec 29, 5:44 am, "Rajeev Kumar" <er.rajeevdo...(a)rediff.com> wrote:
> Respected Sir,
>  I am doing My M. tech project .I am new to matlab .I am designing feed forward backpropagation network through. I have 1 input and hundred 100 outputs. When I enter the data in matrix form e.g.   Inputs=[1 2 3 4 5];
> And t=[ 0   0.2  0.3&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#­8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;­&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;up to 100 columns;
>               &#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#823­0;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8­230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&­#8230; up to 100 columns;
>              &#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#823­0;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8­230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&­#8230;&#8230; up to 100 columns;
>              &#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#823­0;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8­230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&­#8230;&#8230;&#8230;up to 100 columns;
>             &#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#823­0;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8­230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&#8230;&­#8230;&#8230;&#8230;&#8230; up to 100 columns;];
> Means target matrix is 5*100 i.e 5 rows and hundred columns .but the matlab gives error i.e
> ??? Error using ==> network.train at 228
> Targets are incorrectly sized for network.
> Matrix must have 100 rows.
> Please help me how can  size my data for input and output targets. Please reply me. My email id is er.rajeevdo...(a)rediff.com   . I will be very greatful to you.
> Thanking You.

If you have N observations of I input variables and O output
variables,
then

size(p) = [ I N ]
size(t) = [ O N ]

Hope this helps.

Greg

P.S. See my post on pretraining advice

 | 
Pages: 1
Prev: get intensity from pixel impixelinfoval.m
Next: save Variables in Gui view