GENIUS FORUMS

ataxas wrote:

Pincho Paxton wrote:I was talking about the first Neural Networks which used Yes/No/Maybe as the rules. The computer programes that were taught to speak in English.

Either I am still not understanding your meaning correctly, or you are mistaken. Neural networks, even at their humble beginnings, used much more complex internal variables than "Yes" "No" "Maybe" and back-propagation rules to modify the connection weights between nodes.

Here is a link for the reference of others, who may not be entirely familiar with the debate: http://en.wikipedia.org/wiki/Artificial_neural_network - A link showing neural nets as they are today.

http://en.wikipedia.org/wiki/Feedforward_neural_network - A link showing the simplest and earliest neural net, the single-layer perceptron, and the second generation, a multilayer perceptron, which is the first type to use back-propagation techniques.

The second link, I think, clearly shows that the earliest type of neural net uses analogue internal values (decimal points between -1 and 1), and that it was not used to model English, as, in the 1950s when it was devised, there was no hardware capable of supporting such a function.

In double-checking my information, I had the thought that you might be referring to the first real-world application of a neural network. It was a project named MADALINE, which is still in use, filtering static out of phone lines.

https://www.thedacs.com/techs/neural/ - A link to an article about the history of neural networks, containing information about MADALINE. The history of neural nets is in part 3.0; the link will not link directly.

I then set about searching for the model to which you refer, which seems to be this model:

http://www.springerlink.com/content/n09gg0v2x5736887/ - A summary of the paper in question.

The paper was published in 1994, and submitted in 1993, almost a half-century after the first physically constructed model of a neural net. It is disquieting to accuse another of intellectual dishonesty, and so I will assume that you were convinced of the accuracy of your information, and so did not check your facts.

Pincho Paxton wrote:Biological brain is sentient, and you will not find the sentience, but you might be able to use it some day. Sentience is most likely analogue, electrons could be analogue, but I would say that the move through a aether, and that would make them work in steps, although, very, very tiny steps, totally indivisable.

I wonder what you mean by electrons being analogue. Do you mean their movement? What do you mean when you speak of "aether"? Something like this?:

http://scienceworld.wolfram.com/physics/Ether.html

Then, do you deny the existence of vacuums?

Nothing can move totally analogue because you cannot infinately half the distance of a movement. So the smallest particle cannot move half the distance of the aether, else the aether would need another aether, and that is a Paradox. So there are tiny steps to everything in the physical universe. Some particles have to be totally reconstructed at a distance.

I deffinately deny the existence of vacuums, impossible to see through space in a vacuum. The photons would fly apart in every direction. So the Moon would just scatter all over the place into a mist. Colours would also be impossible, and waves would never propogate. Nothing would be able to move without an Aether, because nothing would be there to pass the message of movement along.

The paper was published in 1994, and submitted in 1993, almost a half-century after the first physically constructed model of a neural net. It is disquieting to accuse another of intellectual dishonesty, and so I will assume that you were convinced of the accuracy of your information, and so did not check your facts.

I never knew about other neural networks before the one that I mentioned which was shown on TV. But during that program they said that this was the first Neural Network to be modelled from the human brain, and that the scientists were surprised that three simple rules like YES/NO/MAYBE would actually work like a human brain. I think it was Horizon.

Pincho Paxton wrote:Nothing can move totally analogue because you cannot infinately half the distance of a movement. So the smallest particle cannot move half the distance of the aether, else the aether would need another aether, and that is a Paradox. So there are tiny steps to everything in the physical universe. Some particles have to be totally reconstructed at a distance.

You should read a summary of Planck's work. It may not convince you, but I think to have a well-formulated opinion, one must research the opposing opinion.

Pincho Paxton wrote:I deffinately deny the existence of vacuums, impossible to see through space in a vacuum. The photons would fly apart in every direction. So the Moon would just scatter all over the place into a mist. Colours would also be impossible, and waves would never propogate. Nothing would be able to move without an Aether, because nothing would be there to pass the message of movement along.

Nothing needs to pass the message along. The message passes itself along. A photon experiences no friction, because it is massless energy.

Pincho Paxton wrote:I never knew about other neural networks before the one that I mentioned which was shown on TV. But during that program they said that this was the first Neural Network to be modelled from the human brain, and that the scientists were surprised that three simple rules like YES/NO/MAYBE would actually work like a human brain. I think it was Horizon.

Horizon stretches the truth to make it more sensational.

I don't want to appear as though I am scolding you; that would be ridiculous of me. However, I am reasonably assured that Horizon, and pop-scientific programmes in general, are no solid sources of information. I also hasten to add, that simply because one has not heard of a thing, it does not mean that one cannot find it if bothering to look.

ataxas wrote:

Pincho Paxton wrote:Nothing can move totally analogue because you cannot infinately half the distance of a movement. So the smallest particle cannot move half the distance of the aether, else the aether would need another aether, and that is a Paradox. So there are tiny steps to everything in the physical universe. Some particles have to be totally reconstructed at a distance.

You should read a summary of Planck's work. It may not convince you, but I think to have a well-formulated opinion, one must research the opposing opinion.

Pincho Paxton wrote:I deffinately deny the existence of vacuums, impossible to see through space in a vacuum. The photons would fly apart in every direction. So the Moon would just scatter all over the place into a mist. Colours would also be impossible, and waves would never propogate. Nothing would be able to move without an Aether, because nothing would be there to pass the message of movement along.

Nothing needs to pass the message along. The message passes itself along. A photon experiences no friction, because it is massless energy.

Pincho Paxton wrote:I never knew about other neural networks before the one that I mentioned which was shown on TV. But during that program they said that this was the first Neural Network to be modelled from the human brain, and that the scientists were surprised that three simple rules like YES/NO/MAYBE would actually work like a human brain. I think it was Horizon.

Horizon stretches the truth to make it more sensational.

I don't want to appear as though I am scolding you; that would be ridiculous of me. However, I am reasonably assured that Horizon, and pop-scientific programmes in general, are no solid sources of information. I also hasten to add, that simply because one has not heard of a thing, it does not mean that one cannot find it if bothering to look.

You make too many quotes taken from the nonsensical science that I have spent years avoiding. A message passes itself along does not make sense. A photon experiences no friction is the reverse of what matters. Scolding me would be difficult being as you are just quoting information that you have remembered, rather than constructing the actual data yourself. I have no idea if the programs that you mentioned were true neural networks, but you seem to be overly concerned with timeline, rather than the creation of artificial intelligence. Finding things on the net is not really as important as working towards the future.

Pincho Paxton wrote:You make too many quotes taken from the nonsensical science that I have spent years avoiding. A message passes itself along does not make sense. A photon experiences no friction is the reverse of what matters. Scolding me would be difficult being as you are just quoting information that you have remembered, rather than constructing the actual data yourself. I have no idea if the programs that you mentioned were true neural networks, but you seem to be overly concerned with timeline, rather than the creation of artificial intelligence. Finding things on the net is not really as important as working towards the future.

I am unsure of what you are trying to say in this post. If your position is that you cannot accept the research of others without investigating their work personally, I would agree with you, although I might not go so far as to wish to operate a particle accelerator myself. I do not think it presumptuous to read a convincing paper or two, that detail the same opinion and its corroborating evidence, and be satisfied, at least until contradictory evidence is made public. What worries me about your post is the disregard for the work of others, which shows you have not reasearched the subject to my satisfaction, if I am considering the validity of your opinion. It is one thing to remain unconvinced by a theory at large (I am in no way suggesting that it is wise to simply conform without thought), however, quite another to dismiss such theories out-of-hand as nonsensical, when to be justifiably dismissive would require being familiar with the claims associated with them.

The problem with ether is that if it is real, it is completely unreactive, does not respond to ionisation, radiation, gravity, magnetic fields - any of the super-atomic forces or influences - if it exists, we cannot know it exists, and there is no positive evidence for its existence. Does ether consist of any particles we know of? They must be neutral and massless; neutrinos, perhaps? How on earth would neutrinos "carry" photons?

As to your dismissal of every neural net before 1993, simply because they were created before the one you saw on a TV programme, I must ask: what is it that gives this particular neural net such precedence? Is it because you saw it on television? Or is there something else that makes you certain that this program was a neural net, and the ones before were not? To shed some light on the subject: neural networks have made a gradual progression from the very simplest construction to those complex enough to model human speech, learn how to drive (on a limited stretch of road, unfortunately), recognise male and female faces as such without direct instruction, and so forth; they operate on the same basic premise, and are thus all neural nets. My expertise on the subject is, sadly, limited to one who uses and configures simple neural nets for computational linguistic applications - a hobby, so to speak - but I am certain of my statements.

ataxas wrote:

Pincho Paxton wrote:You make too many quotes taken from the nonsensical science that I have spent years avoiding. A message passes itself along does not make sense. A photon experiences no friction is the reverse of what matters. Scolding me would be difficult being as you are just quoting information that you have remembered, rather than constructing the actual data yourself. I have no idea if the programs that you mentioned were true neural networks, but you seem to be overly concerned with timeline, rather than the creation of artificial intelligence. Finding things on the net is not really as important as working towards the future.

I am unsure of what you are trying to say in this post. If your position is that you cannot accept the research of others without investigating their work personally, I would agree with you, although I might not go so far as to wish to operate a particle accelerator myself. I do not think it presumptuous to read a convincing paper or two, that detail the same opinion and its corroborating evidence, and be satisfied, at least until contradictory evidence is made public. What worries me about your post is the disregard for the work of others, which shows you have not reasearched the subject to my satisfaction, if I am considering the validity of your opinion. It is one thing to remain unconvinced by a theory at large (I am in no way suggesting that it is wise to simply conform without thought), however, quite another to dismiss such theories out-of-hand as nonsensical, when to be justifiably dismissive would require being familiar with the claims associated with them.

The problem with ether is that if it is real, it is completely unreactive, does not respond to ionisation, radiation, gravity, magnetic fields - any of the super-atomic forces or influences - if it exists, we cannot know it exists, and there is no positive evidence for its existence. Does ether consist of any particles we know of? They must be neutral and massless; neutrinos, perhaps? How on earth would neutrinos "carry" photons?

As to your dismissal of every neural net before 1993, simply because they were created before the one you saw on a TV programme, I must ask: what is it that gives this particular neural net such precedence? Is it because you saw it on television? Or is there something else that makes you certain that this program was a neural net, and the ones before were not? To shed some light on the subject: neural networks have made a gradual progression from the very simplest construction to those complex enough to model human speech, learn how to drive (on a limited stretch of road, unfortunately), recognise male and female faces as such without direct instruction, and so forth; they operate on the same basic premise, and are thus all neural nets. My expertise on the subject is, sadly, limited to one who uses and configures simple neural nets for computational linguistic applications - a hobby, so to speak - but I am certain of my statements.

You seem to be talking about papers written about neural networks, and I am taking about AI which is computer generated neural networks. The time difference that you talk about is irrelivent because I am only concentrating on this thread which is about AI, and that is computer generated. I don't know why you think that written papers are applicable to my posts. A brain is a machine, and a book is hard to test like a machine, but a computer program is very comparable. So the 90's were the most relevent to this thread.

You are incorrect in assuming that the first neural network modeled on a computer was done so in the 90's.

As the Wikipedia article I linked says, the first neural network to be physically instantiated was a perceptron, in the 1950's.

From 2004, more evidence that AI won't come from wisdom:

Scientists have grown a brain in a petri dish and taught it to pilot an F-22 jet simulator