A quick and free library for MT4, much to the delight of neuralnetworkers - page 2

New comment
 
Graff >> :


Did you try to feed different currency pairs' prices on the NS input? If yes - what will be the result?

I'm about to drop everything and start trying to feed shit to the inputs.


If you have to, then do it. You do not have to inform about the results, they are known in advance.

 
joo >> :

http://www.codeproject.com/ - all tastes and colours

Are you not confused about anything? The known places, including this one, I've looked at. There are no recurrence grids. If my search fails me, please point me to the URL where it is located.

 

Loaded the library as instructed. I am trying to run Expert Advisor with network off - I got a message in my log:

20:25:44 2009.11.17 21:35 NeuroMACD GBPUSD,M5: dll calls are not allowed; 'Fann2MQL.dll'-'f2M_parallel_init

Help.

 
sergkr1 >> :

Loaded the library as instructed. I am trying to run Expert Advisor with network off - I got a message in my log:

20:25:44 2009.11.17 21:35 NeuroMACD GBPUSD,M5: dll calls are not allowed; 'Fann2MQL.dll'-'f2M_parallel_init

>> Help.

Turn on possibility of working with dll.

Menu: "Tools" - "Options" - "Expert Advisors".



Good luck.

 
VladislavVG >> :

Enable the possibility of working with dll.

Menu: "Tools" - "Settings" - "Expert Advisors".


>> Good luck.



Thank you - it's working!

 
Graff писал(а) >>

Have you tried to send prices of different currency pairs to the input of NS? If yes - what will be the result?

It's a no-brainer to submit prices, full and fair. Price increment is an option, but it's also a lame option.

----

But the library really seems worth poking around. I will check it out tomorrow, it's interesting.

 

To the author: Hats off!

There was a long post with test result - disappeared.

You must be mistaken. Number of inputs AnnInpurs=30 and number of neurons AnnsNumber =(AnnInpurs/2+1)=16. I don't know if it affects the result.

I don't quite understand this part of the program:

 /* Создание нейросети */
      ann=
          f2M_create_standard(4, AnnInputs, AnnInputs, AnnInputs/2+1,1);

Это что f2M_create_standard(4,30,30,16,1);??

Количество входов/выходов можно изменять или они зашиты твердо?
Не запускал.Разбираю текст.
Успехов
 

vlad1949 писал(а) >>


Количество входов/выходов можно изменять или они зашиты твердо?

It's not like anyone's forbidding it.


/* Создает стандартную полносвязную нейронную сеть с обратным распространением ошибки (backpropagation neural network)
* num_layers - общее число слоев, включая входной и выходной слои.
* l1num - число нейронов в 1-м слое (входы)
* l2num, l3num, l4num - числа нейронов в скрытых и в выходном слое (в зависимости от числа num_layers).
* Возвращает:
* индекс хендла нейросети, -1 в случае ошибки
*/

int __stdcall f2M_create_standard(unsigned int num_layers, int l1num, int l2num, int l3num, int l4num);
 
Reshetov писал(а) >>

It seems that no one forbids it.

Yuri, in your Expert Advisor, which is attached on the first page, in int start()

double train_output[1];

and later train_output[1]= 1;

Is this a mistake? Or maybe it should be?

 
ANG3110 >> :

Yuri, in your EA, which is attached on the first page, in int start()

double train_output[1];

and later train_output[1]= 1;

Is this a mistake? Or it must be?

It should be train_output[0].

123456789...57
New comment