Thread: يتيح بناء المستغل على الشبكة العصبية معا

hi.... I will testing it... I hope this one to be mature. .

What form of NN are you currently using?

I'm curious to test it ---------------------------------------------
http://www.myfxbook.com/members/pc8multifx

Originally Posted by ;

مرحبًا Neuro ، لدي نظرة خاطفة على شفرة المصدر ولا يبدو أنك تقوم بتطبيع معلومات الإدخال. يبدو أنك تقوم بإطعام الأسعار مباشرة من المجتمع. يجب تطبيع معلومات الشبكة العصبية إما من -1 إلى 0 أو 1.

Agreed. If you're working with information in a range besides [-1;1] or [0], then you have to normalize those data. This is the theory: - You decide the information range. Let say you are working on EURUSD, which means that your data range will probably be something similar to [1.1;1.6] - You convert all data to the range [0;1] (or [-1;1] if needed) - You automatically procedure the information thru the ANN - You de-scale the resulting outputs using the exact same scale. In this instance, you convert all information from range [0;1] back to [1.1; [ 1.6]. Here's a function ported in MQL4 from Atmel-C (C-based language used to program Atmel AVR chips ) that just do this: Inserted Code double map(double x, double in_min, dual in_max, dual out_min, double out_max) return ((x - in_min) * (out_max - out_min)(in_max - in_min) out_min); Example: To convert x from range [0.5;2.5] to range [0;1]: Inserted Code x2 = map(x,0.5,2.5,0,1);