"New Neural" is an Open Source neural network engine project for the MetaTrader 5 platform. - page 87

 
fellow:
..................I need to start with the elementary: the creation of a neuron. Enough theories, mathematical explanations. We need a neuron in MQL5!

I'll write. What do you need, a fast or a universal one? Don't talk rubbish, answer me specifically. Now.

 
yu-sha:

Those who know how to "do" have already done so.For themselves, of course.


echoing the previous statement

Serj_Che:

Who needed to have done everything for themselves. But it all depends on funding, willing to find.


in fact, it may be a question of "who will sell the ready-made NS engine to the meta-quotes for the public".

And if practice does not deceive, then these "who can" and "all for themselves" - there are only two of them yu-sha andSerj_Che.


So guys be brave. name a price for your work. maybe the MC will be interested.

 
MetaDriver go fast, go fast :D
 
sergeev:

... and if practice does not deceive ...

It does :)
 
It's just that no one wants to share their secrets
 
fellow:
No one wants to share their secrets.
People just don't want to deal with other people's cockroaches, they have enough of their own.
 

Offered, once, to head the department (a vacancy expected, and can be temporary), which is responsible for GA in the staff optimizer - in response - silence. At least would have answered something like "you hear, sonny, we have mustache ourselves" but no, the silence.

Well, okay, okay, pride - a sin, no so no. Now again, it seems to offer the money participants, and something says - "Why the hell do you need it? Here's the truth, why? Ideas - a lot, a lot of codes - but for some reason it is not worth ... Why? I do not know ... Either I already think that if something is born, it will be mine and only mine, or...

You know, people, this isn't just an OpenSource project. It's not. It's an engine project that will be a standard component of the platform, a component that allows you to make money. Money, that's the sin that hangs over everyone who visits this resource and no one has the right to judge anyone on this topic. Toad, you see.

 

Okay, you say, but what do you have to say on the subject?

Okay, I will.

Whatever the brain is (either biological neurons or artificial ones) the creature possessing it in nature is not isolated from the outside world (every creature at birth immediately possesses a set of knowledge "sewn" into its proteins).

Has anything been said about this in this thread before? No.

Every researcher who uses the neural net as a descriptive characteristic of the market environment immediately makes the mistake of giving the net a worthless "freedom of choice. That is, the result is doomed in advance, since the "choice" will be conditioned by the learning period. So what's the point? Why does a grid that roughly works on a learning period begin to make predictions over time that are backward-looking to the market? Why? Because there is no relationship to the market!

 
joo:
ASSERT(write ^ booze)
 

What's the way out? There is a way out! It just can't not be there.

Postulate: If a network that works great during the training period works well outside of training, then something is wrong. It's a fact, get over it. What is this mazafucka, what is this vicious bloody circle? However, there is a way out. You can make sure that the network is not the opposite of the trained (and therefore unprofitable). The network should not be the predictive component.

Ok. Let's go back to the brain. There is information that comes to the retina in the form of waves... Now think about what will change if the wavelength changes or something else (and at the same time the retina) - at the output information about the image will change?