Artificial neural networks. - page 12

[Yedelkin]

Who is following the topic: a new article. You can't realise on iron what happens in solution, as long as there are no such approaches.

[Alexey Topounov]

SProgrammer:

Neural networks are primarily designed for pattern recognition (automatic retrieval of previously seen situations), but not for the detection of any patterns.

And if a pattern is transformed (imagined) into an image?

[[Deleted]]

Can someone post the code of elementary neural network, which could learn something, but all words, at least let me see what this thing is, just elementary something .... thank you!

[Igor Makanu]

IvanIvanov:
And can someone to post the code of an elementary neural network, which could train something, because all words, at least let me see what this thing is, just elementary something .... thank you!

if on mql5 then here:

https://www.mql5.com/ru/code/1323

https://www.mql5.com/ru/code/1324

https://www.mql5.com/ru/code/1267

https://www.mql5.com/ru/code/596

at the last link, in the discussion, there is an example of teaching multiplication table to NS

SZZY: search through forum "neural networks" in articles section also much useful

[[Deleted]]

IgorM:

if on mql5 then here:

https://www.mql5.com/ru/code/1323

https://www.mql5.com/ru/code/1324

https://www.mql5.com/ru/code/1267

https://www.mql5.com/ru/code/596

at the last link, in the discussion, there is an example of teaching multiplication table to NS

ZS: search the forum for "neural networks" in the articles section too

Is this the most basic? Pardon the expression.....

[Igor Makanu]

IvanIvanov:Is this the most basic of all? Pardon the expression.....

Well, if you have already had time to read and understand the code in https://www.mql5.com/ru/forum/5010, then for you all elementary.

ZS: Although I suspect that you immediately unsubscribed to my message without trying to figure it out (between posts about 6 minutes), I'm with the 2x2 training NS managed to figure out why they (NS) are needed

[[Deleted]]

IgorM:

Well, if you have already had time to read and understand the code in https://www.mql5.com/ru/forum/5010, then for you all elementary.

ZS: Although I suspect that you immediately responded to my message without trying to figure it out (between posts about 6 minutes), I'm with the 2x2 training NS managed to figure out why they (NS) are needed

Why, ran across the links, but immediately uninspired. for me, this is too high level for a start, on the multiplication table, I want to try, but I'm afraid to mess up, there are three versions of the code with adjustments, could you repeat the final code here or can be in private

[Igor Makanu]

IvanIvanov: Could you repeat the final version of the code here or in person?

this https://www.mql5.com/ru/forum/5010#comment_329221 works correctly, experiment with the code - output all the information you want, if you figure out the "2x2" then it will be easier to understand why we need NS = a universal mathematical apparatus when we do not have a formula for the initial function, but have a fairly large sample of raw data

[[Deleted]]

IgorM:
this https://www.mql5.com/ru/forum/5010#comment_329221 works correctly, experiment with the code - output all the information you are interested in, if you figure out "2x2" then it will be easier to understand why we need NS = a kind of universal matrix when we do not have the formula of the initial function, but we have a fairly large sample of input data

Yeah, ... if through the editor to create it will be a library, I understand correctly from the title of the quoted branch?

[Igor Makanu]

IvanIvanov: Uh-huh, ... if you create it through the editor, what will be the library, I understand correctly from the title of the quoted branch?

yes, the class_NetMLP.mqh library in the folder ....\MQL5\Include

in the meta editor, create a script and copy the code