Hybrid neural networks. - page 10
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Is it allowed to attach ebooks to posts here?
I don't think so, copyright infringement. Better an announcement and then private or email.
I will try it. I will find time for programming (I really hope so :) this weekend. Will it be possible to ask for support, if anything?
What is there to try? If all the algorithms mentioned here have a real life, not just words,
it takes a couple of minutes to insert a formula instead of a fitness function by copy-pasting. :)
Of course you can, we're exchanging experiences here, aren't we?
What's there to try? If all the algorithms you've heard here are real, not just words,
it takes a couple of minutes to insert a formula instead of a fitness function by copy-pasting. :)
You're welcome, we're exchanging experiences here, aren't we?
My perceptron and back propagation algorithm and genetics are all in Excel. I'm learning MQL at a brisk pace, but your function's implementation won't fit my Excel yet. :) That's what confuses me when I start thinking about implementing algorithms in MQL. If and when the need arises to check for retraining and neural amplitude analysis, will MQL be able to help the sufferers. What do you think?How does your algorithm encode parameters?
My perceptron and back propagation algorithm and genetics are all in Excel. I'm studying MQL at a great pace, but your function's realization won't suit my Excel yet. :) That's what confuses me when I start thinking about implementing algorithms in MQL. If and when the need arises to check for retraining and neural amplitude analysis, will MQL be able to help the sufferers. What do you think?
Excel is great, of course, but rightly so, not our method, in terms of best integration with MT is not the best option.
I suspect you're using some kind of off-the-shelf software. It's not bad, but you already have a problem with replacing the fitness feature.
As for checking, analysis and anything else you can think of, everything can be implemented by MQL4.
How does your algorithm encode parameters?
You don't. The parameters are processed directly. This, among other things, increases the speed because there is no encoding-decoding into binary or Grey's code.
You can't. The parameters are processed directly. This increases performance, among other things, because there is no encoding-decoding into binary or Gray's code.
How then does crossing and mutation happen if parameters are processed directly?
Take an accessory function and bend n from min to max to min (max limit).
Ora sigmoidal function from minimum to maximum and from maximum to minimum.
And how then crossing and mutation occur if parameters are processed directly?
If I use the terms "crossing" and "mutation", it is only because I have not invented other terms for the operators of my algorithm. But because of the similarity of approaches I use these terms.
Look here Continuous Genetic Algorithms. It describes a method without coding.