Thread Subject: help required

dear All,

              Greetings of the day!


 I have following doubts in my mind.

1) While training, performance metric is shown in the top
of the graph. What does this signifies.
It is best when high , or low?

2) What is goal?

3) In my openion if am am getting a stright line parallel
to X axis, it means that there is no effect of increasing
the no. of epoche in training.

What does a graph that is, with time comming down to x
axis signifies?

 

I would be thankful if you could spare some valuable time
to answer my quaries.

 

Thanking you in anticipation,

 

regards,

jeetendra

 

 


 

"Jeetendra Pande" <hi_jeetendra@rediffmail.com> wrote in
message <g64bju$9sm$1@fred.mathworks.com>...
> dear All,
>
> Greetings of the day!
>
>
> I have following doubts in my mind.
>
> 1) While training, performance metric is shown in the top
> of the graph. What does this signifies.
> It is best when high , or low?
>
> 2) What is goal?
>
> 3) In my openion if am am getting a stright line parallel
> to X axis, it means that there is no effect of increasing
> the no. of epoche in training.
>
> What does a graph that is, with time comming down to x
> axis signifies?
>
>
>
> I would be thankful if you could spare some valuable time
> to answer my quaries.
>
>
>
> Thanking you in anticipation,
>
>
>
> regards,
>
> jeetendra

You might get a better response if you explained exactly
what it is you are trying to do.

You haven't said what application you are using that has a
graph in the top of the graph. Without more information
your question appears meaningless.

Regards

Dave Robinson

On Jul 22, 6:09=A0am, "Jeetendra Pande" <hi_jeeten...@rediffmail.com>
wrote:
> dear All,
>
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 Greetings of the day!
>
> =A0I have following doubts in my mind.

Please refrain from using this type of title.

You neglected to mention in your title or in the text, that
the topic is neural networks designed via the NN Toolbox.

Therefore, you wasted the time of thousands of good-hearted
people.

> 1) While training, performance metric is shown in the top
> of the graph. What does this signifies.

It depends on which metric you chose.

doc train
help train

will explain the options available and default value.

TRAINLM default is mean-square-error
doc train
help train

TRAINBR default is sum-square-error
doc train
help train

> It is best when high , or low?
>
> 2) What is goal?

The stopping value of the performance metric.
>
> 3) In my openion if am am getting a stright line parallel
> to X axis, it means that there is no effect of increasing
> the no. of epoche in =A0training.

Yes.

> What does a graph that is, with time comming down to x
> axis signifies?

I don't understand the question.

> I would be thankful if you could spare some valuable time
> to answer my quaries.

You are welcome provided the next time you start a thread
you use an informative title.

Hope this helps.

Greg

P.S. you might benefit from my post on "pretraining advice"