Интерполяция, аппроксимация и иже с ними (пакет alglib) - страница 16

[FxTrader562]

Maxim Dmitrievsky :

Hi, yes, I try various ideas that are discussed here https://www.mql5.com/en/forum/86386/page1056

Good to know this ... I thought you just gave up trying any more :))

That is a very long thread, but I will see what is the discussion going on there.

So you're going to be active ONLY in that thread?

Or this thread will also be active?

Also, have you got into this article of SVM:

https://www.mql5.com/en/articles/584

[FxTrader562]

Hi Maxim,

Also, if you need the source code of any of the indicators like Fibonacci levels, TrendLines, Local High Low(Support resistance levels) etc which are discussed in the other thread, then please let me know and I will post all of them here. 

[Maxim Dmitrievsky]

FxTrader562:

I found a good way to transform predictors: Kolmogorov-Gabor polynomial

https://en.wikipedia.org/wiki/Group_method_of_data_handling

[FxTrader562]

Maxim Dmitrievsky:

I found a good way to transform predictors: Kolmogorov-Gabor polynomial

https://en.wikipedia.org/wiki/Group_method_of_data_handling

Great!!

So basically you are trying to implement a similar method like Synergy method of classification as explained by Vapnik in his video. Am I right?

In GMDH do we still need to use RDF?

As per the explanation what I understood is that GMDH itself acts as a neural network and functions very similar to RDF like taking sample of input data and giving output. Am I correct in understanding?

Also, I saw your optimization pictures in the other thread. So is the coding complete or are you still looking to add something to improve the algo?

I am sorry to ask you so many questions. I want to know exactly where you are now in terms of coding so that I can participate and contribute to make the project complete faster:))

[FxTrader562]

HI Maxim, 

I think we should continue our discussion in this thread since there are people who are not happy with our continued posts in the other thread...

Or you can come to private chat in MQL5 so that it will be easy to share code with each other and discuss new ideas and complete the coding and testing much faster..please update me...

[Igor Makanu]

[Igor Makanu]

библиотека alglib из поставки МТ дает возможность с RBF сетями работать? - пролистал инклудники не увидел ((

http://www.alglib.net/interpolation/fastrbf.php

[Maxim Dmitrievsky]

Igor Makanu:

библиотека alglib из поставки МТ дает возможность с RBF сетями работать? - пролистал инклудники не увидел ((

http://www.alglib.net/interpolation/fastrbf.php

не, нету. Есть только то что в русскоязчной версии сайта

даже случайный лес в англ. версии лучше, обновление было

[Igor Makanu]

Maxim Dmitrievsky:

не, нету. Есть только то что в русскоязчной версии сайта

даже случайный лес в англ. версии лучше, обновление было

этот код я видел https://www.mql5.com/ru/code/1267

дочитал книжку по НС, там в главе про сети радиально-базисных функций писали, что вот именно RBF-сети и нужно юзать для аппроксимации, вот в топик и запостил

[Denis Zhivaev]

А что подается на вход в массивы X, Y?

INPUT PARAMETERS:
    X           -   spline nodes, array[0..N-1].
    Y           -   function values, array[0..N-1].