Thread Subject: 1st example of NNet User's Guide does not work

Hello,

I am trying to learn NNet in matlab but the very first
example of matlab does not work on my matlab 2006.

As soon as I enter the "net = newlin(P,T)" command I receive
below error:

>> P = [1 2 2 3; 2 1 3 1];
>> T=[-100 50 100];
>> net=newlin(P,T);
??? Error using ==> newlin
Input ranges is not a two column matrix.

I'll appreciate if someone can help.

Regards,
Mac


Text of the User Guide's sample:
---
Suppose that the network simulation data set consists of Q =
4 concurrent vectors. Concurrent vectors are presented to
the network as a single matrix:

P = [1 2 2 3; 2 1 3 1];

Suppose that this network typically outputs -100, 50 and
100. This is arbitrary for this example. If you were solving
a real problem, you would have actual values.

T = [-100 50 100];

To set up this feedforward network, use the following command:

net = newlin(P,T);

For simplicity assign the weight matrix and bias to be
W=[1 2] and b=[0]. The commands for these assignments are

net.IW{1,1} = [1 2];
net.b{1} = 0;

You can now simulate the network:

A = sim(net,P)
A =
5 4 8 5

On Oct 3, 11:11 am, "Siamak Sarmady" <sarmadys-rem...@remove-
gmail.com> wrote:
> Hello,
>
> I am trying to learn NNet in matlab but the very first
> example of matlab does not work on my matlab 2006.
>
> As soon as I enter the "net = newlin(P,T)" command I receive
> below error:
>
> >> P = [1 2 2 3; 2 1 3 1];
> >> T=[-100 50 100];

Mismatch. You have 4 2-dim input vectors but only 3 1-dim outputs.

[Nvarin Ncases] = size(P)
[Nvarout Ncases] = size(T)

Try

T = [-100 50 50 100];

Hope this helps

Greg

Thank you very much. I did not understand exactly the same
point. (why T has only 3 members)

This error has been repeated in a few examples of the User's
guide. I wish Mathworks will correct it in next release
because at the very first step of learning, people will come
to problem.

- It's regrettable. Even by changing the T to [-100 50 50
100] it will not work.

??? Error using ==> newlin
Input ranges is not a two column matrix.

It needs the newlin() function to changed to newlind() in
addition to the change to T matrix.

- Who has changed this example to its current version. It
has been different in 2002 and 1998 versions (I checked with
other people's version) and the network was initiated by :

net = newlin([1 3;1 3],1);

But in new version of PDF user's guide it has been changed
to current thing.

- Anyone knows how we can ask support to update the User's
guide with a correct information.

Many people use this user's guide to learn Matlab's NNET and
they will have problem in the first example (and also next
few others).

Mac

"Siamak Sarmady" <sarmadys-remove@remove-gmail.com> wrote in message
news:fe2p7a$4p1$1@fred.mathworks.com...
>- It's regrettable. Even by changing the T to [-100 50 50
> 100] it will not work.
>
> ??? Error using ==> newlin
> Input ranges is not a two column matrix.
>
> It needs the newlin() function to changed to newlind() in
> addition to the change to T matrix.
>
> - Who has changed this example to its current version. It
> has been different in 2002 and 1998 versions (I checked with
> other people's version) and the network was initiated by :
>
> net = newlin([1 3;1 3],1);
>
> But in new version of PDF user's guide it has been changed
> to current thing.
>
> - Anyone knows how we can ask support to update the User's
> guide with a correct information.
>
> Many people use this user's guide to learn Matlab's NNET and
> they will have problem in the first example (and also next
> few others).
>
> Mac

Hi,

To address your question about how to ask support to update the User's Guide,
you can email doc@mathworks.com. If you're using the latest version of the
product, you will also notice that every page of the HTML doc has a link at the
top and bottom that says, "Provide feedback about this page". If you have a
MathWorks Account, then another way to provide feedback is to use the "Request
an Enhancement, or Report a Bug or Documentation Issue" link on the Contact Us
page of the MathWorks Web site
(http://www.mathworks.com/company/aboutus/contact_us/). In this last case, you
will need to log into your MathWorks Account before you can submit the feedback.

FYI, when I saw this newsgroup thread yesterday, I entered a change request in
our tracking system so that someone would look into it. (I work in a different
product area and don't know whether the Neural Network Toolbox documentation
still has the problem you mentioned, or other manifestations of the same
problem.) We definitely appreciate your pointing out the problem because
examples are an important way for people to learn how to use products!

Thanks,
Amanda

Hello

After 4-5 months I visited online matlab documentation and I
see nothing has changed and the example is not corrected.

I had even opened a ticket 5 months ago and asked by email.

http://www.mathworks.com/access/helpdesk/help/toolbox/nnet/

You look for yourself how much commitment has been put on this.

Thanks,
Mac


> Hi,
>
> To address your question about how to ask support to
update the User's Guide,
> you can email doc@mathworks.com. If you're using the
latest version of the
> product, you will also notice that every page of the HTML
doc has a link at the
> top and bottom that says, "Provide feedback about this
page". If you have a
> MathWorks Account, then another way to provide feedback is
to use the "Request
> an Enhancement, or Report a Bug or Documentation Issue"
link on the Contact Us
> page of the MathWorks Web site
> (http://www.mathworks.com/company/aboutus/contact_us/). In
this last case, you
> will need to log into your MathWorks Account before you
can submit the feedback.
>
> FYI, when I saw this newsgroup thread yesterday, I entered
a change request in
> our tracking system so that someone would look into it. (I
work in a different
> product area and don't know whether the Neural Network
Toolbox documentation
> still has the problem you mentioned, or other
manifestations of the same
> problem.) We definitely appreciate your pointing out the
problem because
> examples are an important way for people to learn how to
use products!
>
> Thanks,
> Amanda
>
>

"Siamak Sarmady" <sarmadys-remove@remove-gmail.com> wrote in message
news:fq9apb$hsb$1@fred.mathworks.com...
> Hello
>
> After 4-5 months I visited online matlab documentation and I
> see nothing has changed and the example is not corrected.
>
> I had even opened a ticket 5 months ago and asked by email.
>
> http://www.mathworks.com/access/helpdesk/help/toolbox/nnet/
>
> You look for yourself how much commitment has been put on this.
>
> Thanks,
> Mac
>
>
>> Hi,
>>
>> To address your question about how to ask support to
> update the User's Guide,
>> you can email doc@mathworks.com. If you're using the
> latest version of the
>> product, you will also notice that every page of the HTML
> doc has a link at the
>> top and bottom that says, "Provide feedback about this
> page". If you have a
>> MathWorks Account, then another way to provide feedback is
> to use the "Request
>> an Enhancement, or Report a Bug or Documentation Issue"
> link on the Contact Us
>> page of the MathWorks Web site
>> (http://www.mathworks.com/company/aboutus/contact_us/). In
> this last case, you
>> will need to log into your MathWorks Account before you
> can submit the feedback.
>>
>> FYI, when I saw this newsgroup thread yesterday, I entered
> a change request in
>> our tracking system so that someone would look into it. (I
> work in a different
>> product area and don't know whether the Neural Network
> Toolbox documentation
>> still has the problem you mentioned, or other
> manifestations of the same
>> problem.) We definitely appreciate your pointing out the
> problem because
>> examples are an important way for people to learn how to
> use products!
>>
>> Thanks,
>> Amanda
>>
>>

Hi,

I checked on the status of the change request and it is still open. Sorry for
the inconvenience this has caused you.

By the way, most MathWorks products have updates on a 6-month release cycle.
http://www.mathworks.com/products/new_products/release_model.html
This means that it is possible for a mistake in the documentation to have been
fixed in preparation for an upcoming release, but for the fix to appear on the
Web only a few months later. In your case, you reported the mistake very shortly
after R2007b was released, and R2008a has not yet been released so the
documentation page you're looking at on the Web has not been updated.

R2008a will contain many bug fixes that reflect the commitment of MathWorks to
producing quality software (including quality documentation). At the same time,
I understand that if a particular bug that affects *your* work does not get
fixed, then you may not care about bug fixes that don't relate to your work. I
hope that you'll try not to extrapolate from a single example of a bug, and that
you'll continue to report problems that you might see. An undiscovered and
unreported bug has little chance of being fixed.

Best wishes,
Amanda