Neural networks, how to master them, where to start? - page 15

New comment
 

TheXpert just doesn't understand that universality in networks is nonsense...


A universal network cannot be equally good at recognising pictures, separating into classes, predicting time series, predicting diseases, etc., etc.


That's why there are so many implementations... and so in one particular area, a singleton network sharpened for a specific task in specific conditions will be better than a generic package...


But this in principle comes with experience... :)

 
danja >> :
Where do you get NS2 with the cure?

>> on a spider.

 
LeoV >> :

Weird.... I thought you wrote about this..... I was just wondering what kind of PROF package is this? ....

I don't remember the brand. One of the leading ones in the area a few years ago. The pro version cost $15000.

Solver.it >> :

1. Strange as it may seem to you pirates, but I have all my software licensed... :)

Okay, you got me on that side. Except I'm not a pirate.

2. If you think your own thoughts are bad, what are we talking about?

I just wonder if you do anything yourself or if others always think for you... (those whose thoughts are the ultimate truth) ?? :)

Jeez, where did I say I thought my thoughts were bad?

It's very funny how you manipulate concepts.

It's hard to keep up with you in this business, you are so far from twisting my words as you are showing me here.

 
Solver.it >> :

TheXpert just doesn't realise that universality in networks is nonsense...

This isn't funny anymore. RTFM, this conversation is over.
 
TheXpert wrote >>

Угу, свои в первую очередь не являются.

Didn't you write that? And doesn't it say HIS thoughts (and yours too, to you they are your own...)?


TheXpert >>:
That's not funny. RTFM, this conversation is over.

You find it funny? A lot? I don't.

That's the kind of guy who plants regular corn above the Arctic Circle... because it's universal... :)


And the worst thing is that they teach it to young people... it's not funny really...


RTFM is a great thing...

if you show me WHERE in RTFM it says that neural networks are better to use universal algorithms

and not to improve them, I'll apologise... :)


P.S. And on the subject of "This conversation is over"...

when a person has no constructive arguments, the easiest thing to do is to slam the door... :)

 
Or you can build a neural network using this simple algorithm: if the range of a candle is greater than the average range of the last five candles by a certain coefficient (if the candle is rising, buy, if the candle is falling, sell).
 
If it is not a secret, what is the purpose of the network in your case? After all, you have prescribed the algorithm in detail - go ahead and build an MTS on top of it!
 

Good afternoon.

There was a link in this thread to www.neuroproject.ru. There is a seminar there.

Can any of the experts comment on the programme of this seminar? Is it worth going there for a beginner?

http://www.neuroproject.ru/kpk.php

 
Neutron >> :
If it is not a secret, what for is the network needed in your case? After all, you've written out the algorithm in details - go ahead and build an MTS on it!

I want to learn how to write EAs with a neural network.

>> Is it possible or not?

 

You can, of course!

Only she will learn your rule in 10 lessons and then what?

1...8910111213141516171819
New comment
Reason: