Neural networks. Questions from the experts. - page 24

New comment
 
goga: I think you need more than one network, depending on machine capacity, but all of them should be interconnected even though they will be responsible for different processes


Like a crowd of traders at the stock exchange? Maybe it's easier to hire a crowd of traders than a crowd of neural networks? )))

 
goga:

I think you need more than one network, depending on machine capacity, but they should all be interconnected even though they will be responsible for different processes

I think you need to clear your head about neural networks.
 
LeoV:


Like a crowd of traders at the stock exchange? Maybe it's easier to hire a crowd of traders than a crowd of neural networks? )))

traders have to be paid. it's not profitable. plus the human factor.
 
goga: traders have to be paid. it's not profitable. plus the human factor.
So with neural networks there is not less headache, but more. At least traders understand human language - it is possible to negotiate..... "Everyone has his own shortcomings" .....)))
 
TheXpert:
I think you need to clear your head about neural networks.
But as I understand it, one strategy will sooner or later leak, and everything in life is interconnected. I have divided the project into blocks to avoid confusion. 3 blocks are still being worked on, but they have nothing to do with neural networks yet
 
LeoV:
So with neural networks it's not less of a headache, it's more. At least traders understand human language - we can negotiate..... "Everyone has a disadvantage" .....)))
As for the neural networks, at least traders understand human language - you can come to an agreement. "Everyone has his own weaknesses". People tend to pull in their direction and be lazy. if you really think about using human potential, you need to put all people through a psychologist to find unacceptable defects.
 
goga: If we seriously imagine that we have decided to use human potential in this respect, we need to run all the people through a psychologist to identify unacceptable deviations. then talented people won't drop out, as they will have to work in shifts and need management staff, as the research and monitoring area is quite large, we need a coordination department. plus, someone took a dump, someone didn't notice and the resulting problems and costs are higher
Neural networks are prone to overlearning and overfitting. If you seriously imagine that you have decided to use neural networks in this plan, then.......... it would take a couple of books in 4 volumes each....))))
 
goga:
They will have to work in shifts and need management staff, as the study and observation zone is large enough, they need a coordination department.

BOO-GA-GA-GA-GA. :-)))
Put that (and others like it) nonsense out of your mind. Before you play with NS committees, you need to sort out the CASH... :-))), then their necessity will disappear by itself... :-)), namely, get acquainted with their (NS) basic concepts and purpose "A.Ezhov,S.Shumsky.Neurocomputing and its applications in economics.djvu", Kruglov Fuzzy Logic and Artificial Neural Networks.djvu, F.Wasserman.Neurocomputer Science - Theory and Practice.doc. Besides, here is an article on FANN NS, search the forum - there are specialized branches on these questions - for a start - also familiarize with their contents...

In the trailer is part of my archive on NS - experts, forum excerpts, etc. - Look it over, have a look, if you're interested...

Files:
neuro.zip  608 kb
 
LeoV:
Neural networks are prone to over-learning and over-fitting. If you seriously imagine deciding to use neural networks for this purpose, then.......... it would take a couple of books in 4 volumes each....))))
I'm not saying it's a matter of five minutes or that it's easy. I want to know if it's possible.
 
goga:
I'm not saying it's five minutes and that it's easy. I want to know if it's possible.


Reread your posts above, on this page. Drop the whole NS COMMITTEE - you'll only kill time and you'll certainly gain experience (HAPPY)... :-))) Start small... - See my post above.

P.S. Don't forget to row (shovel) the CASH... :-)))

1...171819202122232425
New comment
Reason: