Article: Price forecasting with neural networks - page 5
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
Mak wrote (a):
15 billionOur brains contain about 100 million neurons (correct me if I'm not mistaken).
3. e.g. the notorious Curve fitting.
Otherwise, just remembering patterns (which don't have to work).
There are standard procedures to see if memorisation or proper learning with generalisation has occurred. And you know them, Mak. But for some reason you don't want to talk about it.
Set up an experiment:
- Create a neural network of 50-100 million neurons.
- Train it for 10 to 20 years.
- And you'll get a pathetic semblance of Homo Sapience... :))
If you think that a modern computer is much faster than your brain, you are seriously mistaken.
A computer keeps one thread of calculations (let it be two today - not the point).
Your brain can run thousands and millions of threads in parallel ...
(it's called subconscious, associative thinking ...)
I've argued something similar with you before, Mak. I recommend you again: read Peters. Both books and carefully. I'm reading the first one now. And I can see how profound he is. It says a lot about memory. By the way, machines have learned how to beat people at chess; think of Kasparych, just recently. Very interesting. By the way, this is one of the most unexpected consequences of our inadequate understanding of the way the brain works...
njel 26.06.2007 00:33
There are about 10-15 three billion neurons in the brain. Each one has an average of 10,000 synapses. And this whole bunch controls the work of the whole body. i.e. creating a sales network does not need information from foot and ear receptors at all. and such nonsense.
When I was young, I foolishly created a network with 50 million neurons. Not 10,000 connections per neuron, but still. And I trained that network for a week. The problem was that I didn't want much from it, and my mini-brain cheekily memorized the training sample. Eventually I got to reurent networks... etc.
In short, it's not about creating an analogue of a trained human brain. You need to create a brain that helps you make decisions. Or if you're confident enough, trade. This is a completely different task.
Well, it's a sensible idea at last.
I don't want to read Peters, I have a mustache myself :)
I don't have the time, and my head's already full...
Machines have learned how to beat people at chess?
Did they really learn?
Or were they programmed to do so?
The fact that a machine can solve stupid combinatorial problems faster than a human does not surprise anyone.
These machines were built for such tasks.
Anyway, no matter how you look at it, NS is just a function of a certain type from a great many parameters.
Training of NS is optimization of some function according to this set of parameters.
Will anyone object?
Google tells me there are 100 billion neurons in the brain.
I don't want to read Peters, I have a mustache myself :)
I don't have time, and my head's already full...
Machines have learned how to beat people at chess?
Did they really learn?
Or were they programmed to do so?
The fact that machines can solve stupid combinatorial problems faster than humans comes as no surprise to anyone.
These machines were built for such tasks.
Anyway, no matter how you look at it, NS is just a function of a certain type from a great many parameters.
Training of NS is optimization of some function according to this set of parameters.
Will anyone object?
No objections.
Simply, a trained neural network executes a linear program as no sane programmer would write it :).
Did they really learn?
Or were they programmed to do so?
The fact that machines can solve stupid combinatorial problems faster than humans doesn't surprise anyone.
These machines were built for such tasks.
No, you don't get it. Stupid overkill is only the first stage in the development of such machines. They were taught to beat grandmasters by Bronstein, the grandmaster himself. If you are not surprised, then you yourself simply do not understand the complexity of the problem.
Did they really learn it themselves?
Or have they been programmed to do so?
The fact that a machine can solve dumb combinatorial problems faster than a human isn't a surprise to anyone.
These machines were built for such tasks.
No, you don't get it. Stupid overkill is only the first stage in the development of such machines. They were taught to beat grandmasters by Bronstein, the grandmaster himself. If it doesn't surprise you, it means that you yourself simply don't understand the complexity of the problem.
I started getting interested in Artificial Intelligence when the topic was still in vogue, and when it seemed that in another year or two, we would start communicating with machines like with humans. A lot has happened since then. AI is no longer a trendy topic. People have realised that everything is much more complicated than it seemed at first.
Yes, there are a lot of heuristics put into chess programs besides simple brute force.
I've even got a book somewhere by Bronstein, with a description of his ideas about this program.
He defended either a candidate's or doctoral thesis on the subject.
Back in Soviet times.
But this is not learning and self-learning.
Playing with this program - it is almost like playing with Bronstein, equipped with a powerful amplifier (the iron for search options).
Once again, machines HAVE NOT LEARNED how to beat humans at chess.
As long as the machines are dumb iron that does what they're put into it.
(I've been familiar with this topic for a long time too - I've been dealing with computers all my life, for over 30 years now)
Can anyone here give, give a clear definition of what intelligence is?
An analysis of the nature of perception. It is by perception that our consciousness differs from artificial intelligence.
Next, still on topic :)
http://www.cirota.ru/forum/view.php?subj=68034
How easy it is to make an atheist, a non-atheist - but. For atheists too )
Atheism has its basis in science, and a logical approach to the phenomena around us. The atheist believes that every object can be described by a set of rules and properties - which will uniquely determine the behaviour of an object.
The apotheosis of this conception in science is the approach of the law of "One interaction". Which unites all physical laws (Gravitation, electromagnetic interactions, nuclear forces - strong and weak interactions).
Moreover, it has already been partially realized that electromagnetic and weak interactions are united by a single law. As a matter of fact, I myself believe the same. An indirect confirmation of this is that at the moment of big bang, when distance is small, formulas of all known laws give practically identical results, in value by vector, etc.
It should also be noted that the scientific approach completely denies human freedom in its intuitive sense. As it completely defines the behavior of objects from and to.
There is also a provision for pure randomness - but it, if any, only limits human freedom. It is clear that freedom of choice is restricted, while if it acts on the level of free will then (when I was a communist, I became a democrat by chance.
A Conscious choice is necessary.
To a reasonably thinking person it should seem that the scientific approach really pretends to a complete and sufficient description of existence. Here if there were something that in principle cannot be described by science and it is proved that it cannot, it would undermine all the basis of atheism. Which I will do below.
Let's turn to what every human being has - consciousness. Which can be subdivided, like the Holy Trinity, into three components - Will, Sense, Reason.
Let us limit ourselves and consider only feelings in a generalized way. This is perception. Everyone can be convinced that he/she possesses it.
I will specify at once that the scientific approach, a logical approach has no pretensions and fundamentally cannot explain the nature of perception.
Let's take a scientific picture of the world and we shall remove perception from it. Will anything change? No. Everything will remain the same. The sun will still shine, the planets will rotate and so on. Even people will have to stay the same, even communicate.
But it will be robots, programs, etc.
Molecules, eletrons, kraks in the end will rotate and fluctuate in their orbits with Doppler correction in the same way. We do not abolish the laws of the universe. Therefore a consistent materialist must claim that perception does not exist. But he cannot assert that. Try to argue that it is an illusion?
But the whole idea is that there shouldn't even be an illusion.
------------------------------------------------------------------------------------
Another argument. If parts don't have some property in principle, even if in embryo, then the system of these parts will not have this property. A very good example:
Take a motor that moves a car. It is stationary and the car is moving.
The point is that its parts can in principle change position in space, so their vectors add up and it is possible for the car to move. If motor parts couldn't move in space, it would certainly not move the car.
This is saying that if molecules do not have the property of perception, the conglomerate of molecules will not have it either. But the man is a conglomerate of molecules.
And the conglomerate will move according to the same physical laws without losing perception.
-------------------------------------------------------------------------------------
Another important moment of perception. Why you cannot dismiss it so easily.
The point is that perception is an absolute truth in itself. But without its logical meaning.
Of course it may seem that you are on the moon, which in its logical meaning would of course not be true. But the fact that you seem to be on the moon is the truth.
More than that and it's the only thing we have directly.
-------------------------------------------------------------------------------------------
Conclusion, the scientific approach, fundamentally cannot describe perception as such.
Similarly pure logic turns out to be insufficient to describe being as such.
(Freedom is of analogous nature to the nature of perception. - Well, it's a supplement.
Atheism is not even logically grounded.
...
Atheism does not even have a logical basis.
There are 2 differently trained ANNs, we feed the same data sets to their inputs, but get different outputs. Objectively the picture at the input is the same, but "perception" of this picture is different for eachneural network. Supposedly the human brain has much in common with neural networks, and his whole life is nothing but training of this network and change of synaptic scales, and due to the fact that all people live different lives, their neural networks are trained differently, hence everyone will have different subjective perception for the same objective reality.
Perhaps I am misinterpreting the term "perception", so all of the above is nothing more than a personal opinion formed as a result of neglecting the humanities at school and uni.
There is also a definition. Intelligence is the ability to discriminate (distinguish).
Next, still on topic :)
That's the thing, it's off-topic. Your argument is a bit off-topic.
Maybe you should take the concept of "intellect" to pieces and then discuss what is possible and what is not possible to realize programmatically.
Not how many neurons are in the head and how many are needed.