A quick and free library for MT4, much to the delight of neuralnetworkers - page 17
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
I get it, but I don't have this feature for some reason. Thought it was your personal additions to this inluder. Isn't it?
It's an extended inluder, but I got it from codebase.
Checked - my values are different and scattered almost evenly. Here is one of the initialisations :
...........
Just added a generator re-initialisation today - don't know how much of an effect it has.
Your weights are more than correct, but what are the output values of the nets with the same input vector?
The output values are different -- that's nice.
But. The nets are initialised with random weights...., but the same weights every time. Checked.
In other words.
-- ANN directory is clean.
-- I run EA and save the network profiles immediately after initialization (16 networks). I save them to the ANN_1 directory
-- I empty the ANN directory again.
-- run EA again and save the network profiles just after initialization (16 nets).
-- compare profiles from ANN and ANN_1 -- they are the same.
It looks like something is wrong with srand()...
There seems to be something wrong with srand()...
Right - I remembered one "trick", which was used in "prehistoric" times ( about 1991 :) - I forgot about this subtlety). Glanced at the Help - and it advises so. :).
Fixed it, in general. Try this option - should be correct.
Somehow "_2" stuck to the name - just rename it.
For some reason the "_2" is stuck to the title - just rename it.
That's right. Because _1 was already there. ))
Vladislav, just a vicious circle.....
With vgTALib_2.dll the initial initialization was normal, but then the outputs became the same as in FANN2MQL :-(
Next, I don't change anything, I write vgTALib_1.dll in the inline again, the outputs are different. Baffle....
Somewhere in the source library hangs a small, but very significant snot. And it looks like they didn't put it there by accident....
You can get the code and description of the library in the article: Using Neural Networks in MetaTrader
Thanks to the author!
The library has turned out to be working!
Thanks to the author - no doubt, even for a non-working library!
Does the topic-starter have any comments on this problem?
Maybe it's time to change the thread title to "On the woe of neuralnetworkers, fast and free library FOR SLIVE........" ???
Maybe it's time for the moderators to remove the above article from view? So as not to enter into temptation. And comment on.....
Why so radical? There's a little problem.
I'm a bit tired. Trouble at home... I'm out of commission for three days...
There's no such thing as a little problem in trading... The library's been compromised. We'll have to sort it out.
Help!
I'm a bit tired. Trouble at home... I'm out of commission for three days...
There's no such thing as a little problem in trading... The library's been compromised. We need to sort this out.
Help!
int StartTick=GetTickCount();
for (int i = 0; i < AnnsNumber; i++)
{
while(GetTickCount()==StartTick){}
AnnsArray[i] = ann_load (ANN_PATH+AnnFullName+ "." + i + ".net");
StartTick=GetTickCount();
}
int StartTick=GetTickCount();
Thank you. It's working.
The library itself seems to have implemented its own delay when creating nets (about 1000/64 ms), but without the lines you suggested this mechanism doesn't work for some reason...