Neural networks, how to master them, where to start? - page 3

 
http://algolist.ru/ai/ here is some useful information on NS and GA to start with
 

Yury V. Reshetov wrote an article https://www.mql5.com/ru/articles/1447/page3#comments

I read the article, but I don't understand how to implement it for my EA.

Ideally I would like to see first the code of Expert Advisor without nero network and understand how it works. And then I want to attach the neuronet to it step by step.

At least on the same difficulty level as described by Yury V. Reshetov.

 
his advisor uses a linear perceptron.... it has little effect, you have to constantly adjust the primaries
 
neronka is machine intelligence, to understand how it "intelligence: works
first of all, you need a machine,
-without practical experience in generating networks, a single off-the-shelf NN product will be incomprehensible,
and NN books without NN practice are toxic.
Take care of your Brain)))
 
Andrey4-min >> :

Let's try to write a simple neural network together, lay it all out... and the experienced pragmists will give us some tips...


There's nothing to lay out. The maths is kindergarten maths. The main problem is to fine-tune it (make sure that links are properly created, properly worked and weights are properly corrected). Actually at this point, I just stopped. As soon as you write it, you immediately get smarter in this matter. I guarantee it. And then it's up to you. What to feed it.... How to configure it... It's not just a field. It's a planet out there.

 
Andrey4-min >> :

and experienced pragrammers, give us a hint...


You know, if you write it simply, with a well-defined configuration, without showing off, a neural network is very simple from the programming point of view. A neuron is an adder and an activation function. that's it! two lines of code... Or one, whichever you prefer. three-dimensional array. three cycles of forward run - three cycles of backward run. and that's all there is to it.

 
danja писал(а) >>
Hi there! I'm new to the forum. Ran through a forum full of useful stuff:). I want to ask the knowledgeable, I want to penetrate into sense of neural networks, but do not know with what to begin while have the general idea has read article by J.Reshetov, now have desire to develop in this direction. Possible useful literature on this topic, maybe somewhere on the forum is, but I have not looked up? Thanks in advance:)

For a start, read "Callan R. - Basic Concepts of Neural Networks.djvu", very well explained.

Then you can :

"Ezhov A.A., Shumsky S.A. - Neurocomputing and its application in economy and business.djvu".

"Osovsky S. - Neural networks for information processing.djvu".

"Haykin S. - Neural networks full course.djvu"

These books, in my opinion, are the most interesting. And there's a lot of literature, google it.

 
Talex >> :

For a start, read "Callan R. - Basic Concepts of Neural Networks.djvu", very well explained.

Then you can :

"Ezhov A.A., Shumsky S.A. - Neurocomputing and its application in economy and business.djvu".

"Osovsky S. - Neural networks for information processing.djvu".

"Haykin S. - Neural networks full course.djvu"

These books, in my opinion, are the most interesting. There's a lot of literature, google it.

Thank you! That's exactly what I need!

 
arnautov >> :

You have to understand, if you write it simply, with a well-defined configuration, without any fuss, a neural network is very simple from the programming point of view. A neuron is an adder and an activation function. that's it! two lines of code... or one... whichever you prefer. Two lines of code. Or one, whichever you prefer.

If I post the EA code and highlight the coofficients from it, will I need to add two more lines to create the neural network?

 
Talex >> :

For a start, read "Callan R. - Basic Concepts of Neural Networks.djvu", very well explained.

Then you can :

"Ezhov A.A., Shumsky S.A. - Neurocomputing and its application in economy and business.djvu".

"Osovsky S. - Neural networks for information processing.djvu".

"Haykin S. - Neural networks full course.djvu"

These books, in my opinion, are the most interesting. There's a lot of literature, google it.

Thank you

Reason: