infantlab / tracx-web Goto Github PK

RMF - Some major new bug has been introduced somewhere. The code seemed to be working perfectly the other day, but now it no longer is. It no longer even gets the Saffran or Aslin data correct. Any idea what's up? But technically, it looks good and I'm glad we have it on our Website now.

CJMA -
I think the simulator is working fine. If you look at the bottom right Bigram Error over Time graph you will see that word internal bigrams are learned better than boundary-cross bigrams or non-word bigrams

The probem is that the simulator is using a different test criteria to matlab. I'd argue that the simulator one is possibly more appropriate.

In matlab you always pass the hidden representation forwards to next step and output just the single delta for final item.
e.g You test the word ABC by getting the delta for
input1 = Hidden(AB)
input2 = C

Let's call this Delta(Hidden(AB), C)

In the simulator, we take the mean of the all the errors encountered in the word and calculate them according to whether the internal hidden representations were below the criterion

So for ABC, we have two different expressions for early and later in learning

Early on when Delta(A,B) > CRITERION, our error is

0.5 * ( Delta(A, B) + Delta(B , C))

Later when Delta(A,B) < CRITERION,

0.5 * ( Delta(A, B) + Delta(Hidden(AB) , C))

When AB has only just been learned Delta(Hid(AB), C) is quite large. So the part-word delta's can end up being bigger than the word deltas. Though (from the Giroux and Rey simulation result) we know that eventually the latter is smaller.

I guess the question is. how do we show all of this in a transparent graphical fashion in the simulator?

These aren't parameters.

somewhere on the activations graph on left hand side show the delta between inputs and outputs.

Instead plot graphs every single time

one very important thing we need on the simulator is the possibility of running EXACTLY the same run again. This means we have to have some way of storing the random number generator seed. And of inserting it to make the program run a particular way.
So you run the program with random settings. You get a very nice run that illustrates some point (that you want to use during a class, for example). You click: "Save seed for this run" and it saves it to some file.
Next time you run the program you can either just run it normally or call the file with the saved seed in it. That is how I do it in all of my code.

The Forward and Backward TPs simulations do not match the Matlab simulator.

allow local encoding
e.g. a = (0, 0 , 0.. 0, 1) b= (0, 0, 0, .. 1, 0)

or distributed coding where Nth letter is given binary encoding.

http://leadserv.u-bourgogne.fr/~tracx/ is not live anymore.