How can i make a closed loop NarxNet with feedback of predicted values?

Hi, i´m trying to make a closed-loop Narxnet feedbacking future values. I mean, with feedback of (y+1) or (y+5), but i can´t find a way to do it. I supposed that using the feedback delay with a positive number i could manage it, but i´m not sure if this is correct: net = narxnet(0,5,20); %20 neurons, 0 delay input, y+5 feedback? Instead of delaying it with 1:d

Thank you beforehand

Answer 1

[Greg Heath]

 "but once predicted, feedback them"

That is exactly what narxnet does. The default values ID,FD = (1:2,1:2) yield

 1. y(t) = f( x(t-1), x(t-2), y(t-1), y(t-2)) for t >= 3

which can also be interpreted as

 2. y(t+3) = f( x(t+2), x(t+1), y(t+2), y(t+1)) for t>= 0

The nondefault input ID = 0:2 adds x(t) to the RHS of 1

The nondefault input FD = 0:2 adds y(t) to the RHS of 1 for OPENLOOP ONLY!!!

FD = 0 is not allowed for CLOSELOOP. Any attempt to do so will result in either an ERROR, or the command is ignored.

Hope this helps.

Thank you for formally accepting my answer

Greg

[Luis Ignacio Ruiz]

Sorry for my ignorance but what´s RHS?

Thank your for the help, much appreciated.

[Greg Heath]

RightHandSide

Answer 2

[Greg Heath]

Read

 help narxnet
 doc narxnet

See posted examples in both the NEWSGROUP and ANSWERS

 greg narxnet
 greg narxnet tutorial

Reading the latest ones first may help.

[Luis Ignacio Ruiz]

I´ve already been looking for answers in the matlab help and forums. I just found informationa about multi-step ahead prediction, but what i really need to know is if there´s some way I can feeback those predictions before appearing. I mean, if the output is giving y(t), I would like to know if there´s a method to feeback y(t+1), y(t+2), y(t+3)... just like the normal feedback of previous values (y(t-1),y(t-2),..) but with the future ones.

Thank you beforehand

Answer 3

[Greg Heath]

If you find a way to feedback signals before they are created,

 CONTACT ME IMMEDIATELY!!!

 I'LL MAKE US BOTH VERY, VERY, WEALTHY!!!

Greg

[Luis Ignacio Ruiz]

Haha, not before they´re created but once predicted, feeback them. I thought that the solution might be in feebacking the predicted values once we have the outputs "Y" in a new net, but I don´t know if it could be possible.

P.D.: Sorry for my lack of knowlegde about this subject. It´s the first year I´m studying neural networks, but it has a lot of potential and it´s a big part of my master´s degree final project so I´m looking as much as i can into it.