Hybrid neural networks. - page 21

New comment
 
registred >> :


What's the parameter to adjust there? Sigma? How will you adjust it? How do you find the best solution? These are the questions that are not quite clear to me. Another thing is that for MLP only the type of error is an essential parameter, I insist on it nevertheless. Of course, MLP gets stuck in the local minimum and there are methods of fighting it. In any case, optimal solution for MLP may be found not through finding in the global minimum for many tasks. If you have worked out something with PNN, that's very good.



About probabilistic networks: PNN But I'm sure in this case it wasn't classical PNN that was used, judging from Better's statements


"No, the NS provides only one value at the output - the rate direction forecast (f).
It is then interpreted by the trading system. If f is greater than some threshold (A), buy, if f<-A, sell."

Championship 2007

 
please tell me how much faster the algorithm in c# will work compared to mql4!
 
gumgum писал(а) >>
Please tell me how much faster the algorithm in c# will be compared to mql4!

I won't say exactly, but you can expect an acceleration of about 20 times...

 
Now imagine, friends, that by working in forex with nets, one gets so trained to build, train and twist in a general sense that another task. A simpler one, will become as easy as cracking a nut. So train in forex (the most difficult BP to forecast) and go forecast company profits, demand, weather, etc. That's where you'll be a pro.
 
StatBars писал(а) >>

I won't say exactly, but you can expect speedup of about 20 times...

And the neural network inside dll (or other algorithm) with mql4 will also work much faster than "everything is written in mql4"?

 
sak120 писал(а) >>

And neural network inside dll (or other algorithm) with mql4 will also work much faster than "everything is written in mql4"?

Actually I was talking about DLL (there's no difference between separate program or dll). But not everything will run much faster there, the big acceleration is mainly in math operations.

 
StatBars >> :

In fact, I was talking about DLL (there is no difference whether it's a separate program or a dll). But not everything will be much faster there, the big acceleration is mainly in math operations.

I didn't manage to connect DLL written from C# to MQL. I made a wrapper, C dll + C# dll

 
StatBars >> :

I won't say exactly, but you can expect an acceleration of about 20 times...

So if I 2000 trained multiplier to <0.001 two hidden layers 22 and 11 neurons waited about 1 hour. All I have to do is remember C.

 
Stanislav Osovsky ″neural networks for information processing″ article 70. Please explain.
 
c-means algorithm who has a detailed description?
1...141516171819202122
New comment