Neural networks, how to master them, where to start? - page 4
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
If I post the EA code and highlight the co-factors from it, would I need to add two more lines to create the neural network?
No. This neuron is one two lines of code. The network in its dumbest variant is a three-dimensional array, every element of which is calculated with these very two lines. So just write a network, no strings attached. Or just one neuron to start with.
I'm making a multilayer pseptron now (I hope I got it right :))
Multilayer Perspectron - nonsense!
-Download NeuroShell 2 with the "wrong LeoV" avatar and go!!!
Where? Preferably the latest version, Russian language and a vaccine for the treatment. If you have a link, send it to me.A multilayer perseptron is nonsense!
Nonsense.
In many problems, a 4-layer Perspectron shows much better results and convergence.
And in some places a 5-layer one is used. Its middle hidden layer captures intermediate data for additional analysis.
By the way, for that matter, deployed recirculation networks are nothing but a perseptron. And a non-linear RNS deployed is precisely a 5-layer perseptron.
I will remain silent about complex networks (with multiple output layers and complex couplings) based on a perseptron for now.
On many problems the 4-layer perseptron shows much better results and convergence.
What a load of crap. A perseptron is a neural network consisting of only one neuron. And of course it can only be a single layer one.
However, neural networks consisting of several neurons are called multilayer neural networks.
What a load of crap. A perseptron is a neural network consisting of only one neuron. And of course it can only be a single neuron.
But neural networks consisting of several neurons are called multilayer.
Google MLP and you'll see that a lot of information is written about this nonsense.
What you mean is Rosenblatt's perseptron and is about 30 years out of date. The basic principles remain the same, though.
What a load of crap. A perseptron is a neural network consisting of only one neuron. And of course it can only be a single neuron.
But neural networks of several neurons are called multilayer.
I don't care what it's called. What matters is how it works. But my intuition tells me that a network of one neuron is called a neuron :) And even if it's nonsense, it doesn't change anything.
Type MLP into google and see that a lot of information is written about this nonsense.
What you mean is Rosenblatt's perseptron and is about 30 years out of date. Although the basic principles remain the same.
Honestly, I found a lot of stuff on this query, but not on the subject.
I mean a simple linear perseptron like the one Reshetov applied. Rosenblatt created his perseptron as a model of brain functioning), a rather primitive model...
I agree about the basic principles.
You have to call things right, not "I don't care what it's called. What matters is how it works."
You have to call things by their proper names, not "I don't care what it's called. What matters is how it works."
Yes? Well, read it.
http://www.scorcher.ru/neuro/science/perceptron/mem29.htm
>> http://alife.narod.ru/lectures/neural/Neu_ch06.htm
Until you and the authors of these articles agree on terminology, I'll assume they're right. And call things what they call them.
The multilayer perseptron is nonsense!
>> And where? It is desirable that the version was the latest, the Russian language and the vaccine for treatment was available. If you have a link, send me a PM.torrent rules,
The latest version just do not advise, firstly not in Russian, and secondly, a lot of "extra", and this "extra" is needed only after the development of Russian
i.e. vintage NS-2 is best, its training examples.ru, its networks in MQL-4.
torrent rules,
The last version just do not advise, first, not in Russian, and secondly, a lot of "extra", and this "extra" is needed only after the development of Russian
i.e. vintage NS-2 is best, its training examples.ru, its networks in MQL-4.
I see. Thank you. I'll see if I can find the unaffected version.
But do NS's official representatives - NeuroProject.ru - sell a non-Russian version for 26000r?