Artificial Neural Networks

[The Cusp]

Anyone at all familiar with neural networks? It looks like there is various software available which allows you to fool around with your own. Kind of like your own Pokemon, you train it to do certain things based on certain input of stimuli. Stuff like predicting the stock market with whatever factors you want to make relevant, like oil or gas prices, market changes, the scores sports games, anything at all. You could even use online dream journal entries. The beauty about neural nets is the find subtle relations between all those inputs in ways we would never think of. There was one experiment done with a neural net hooked up to a circuit board full of various transistors, capacitors, and various electrical doo dads. The would train it to do various tasks using the components it knew nothing about. It accomplished one of those tasks in a manner that still has electrical engineers baffled to this day, because they can't figure out how it works. According to science, that circuit shouldn't be doing what it does, but it somehow works, and nobody can figure out why or how. Gotta be a web link to that somewhere, but I can't find one at the moment. I have it in a book. So anyone know anything about setting up neural networks? I haven't really looked at any of the option in detail yet. This site looks promising. http://www.neural-forecasting.com/

[Xei]

I'd like to see where you can get a cool program where you can mess around with them easily. Know of any? This is actually what I want to research as a career. In my opinion it's at the top of the list of the greatest remaining scientific mysteries... there is currently pretty much no understanding at all of the very basics of how the neurons of the cerebral cortex work.

[The Cusp]

There is a free downloadable CD here with everything you need to get started. http://www.neural-forecasting.com/free_software_cd.htm

[metcalfracing]

This is interesting... I'm checking it out

[metcalfracing]

haha, only like 1 or 2 of those programs run on 64bit vista...

[Xei]

What's neural forecasting? The only result for it on Google is that site, and that site doesn't explain it.

[The Cusp]

What's neural forecasting? The only result for it on Google is that site, and that site doesn't explain it. Basically using a neural network to predict something like stock market behavior.

[Xei]

Bleh, I'm not particularly interested in that. I'm more interested in how actual cognitive functions work, which nobody has figured out yet.

[12g951ad]

hmm new age crap anyone ?.. . . ?

[Xei]

Yeah you'll find all your new age crap related needs here: http://www.dreamviews.com/community/...splay.php?f=19 Bye bye.

[12g951ad]

Thanks ! but not the point

[Xei]

You don't have a point. Do you not understand some of the big words that are being used here or something? Neural networks are a field of serious scientific study, there is nothing 'new age' about them.

[metcalfracing]

hmm new age crap anyone ?.. . . ? I don't even understand how you can draw a comparison between neural networking and new age hippy crap... Neural networks are just computational models used to drawing comparisons from data... among other things

[DuB]

It's certainly an interesting area of research. There are some who think that theories of cognition are increasingly going to based on computational models. I'm somewhat skeptical, however, as developing these models requires a large amount of training in mathematics and, to a lesser degree, computer programming, and the people who have traditionally addressed these research questions (i.e., cognitive psychologists) tend to receive their mathematical training in statistics and quantitative measurement. The emerging, interdisciplinary field of "cognitive science" seems promising. It will be interesting to see how they approach these questions.

[Xei]

I don't see how else we can figure out how the mind works properly if we don't even know how neural networks work. To be frank, the current knowledge about neuroscience is unimpressive. We know that the brain's made of neurons. That's about it. 'Cognitive psychology' or whatever are never going to get us anywhere. The only way forwards is a brute force cracking of how the brain actually works. Otherwise it's like trying to design a computer or figure out how one works with only the knowledge that 'it's made of transistors'.

[DuB]

Compared to more established sciences, I agree that it is somewhat disappointing to think of how little neuroscience has really given us. I view it as one of the last frontiers of scientific investigation. The relatively recent development of functional brain imaging has opened up a lot of possibilities, and I am hopeful that soon we can begin tackling the bigger questions. I have to say that I am a little confused at your comment, "we don't even know how neural networks work." The entire point of neural networks (at least the way that I understand it - which I admit may be flawed) is that we do understand how they work, and that this fact allows us to learn a thing or two about how the cognitive processes that they emulate work. Whereas we cannot break open the black box that is the human mind and examine how it is functioning (at least not currently, or probably any time soon), we can do just that with neural networks and other computational models. So if we have a computational model that appears to mimic human cognition, we can infer that they must be utilizing similar processes. Is this not the way that you understand neural networks? On a side note: I find the analogy of using "brute force" to understand brain functioning to be very amusing. I have never viewed science as a "brute force" sort of process, but to each his own, I guess.

[Xei]

Well, psychology just strikes me as being science for pussies. To understand the thing properly you've gotta take it down to the fundamental level... And about neural networks; nah, it isn't understood how they perform computations etcetera. With computers we know that it uses binary code to represent numbers, etcetera; we don't know what the 'code' is for the brain. I remember the head of the Blue Brain project saying that the field is waiting for a conceptual breakthrough. Until then it's just groping in the dark... It's something I plan to help crack, once I've got a degree in maths and maybe a master's in neuroscience.

[DuB]

Well, psychology just strikes me as being science for pussies. Heh. Okay, I'll bite... What exactly is it about psychology that strikes you as being, ahem, "for pussies?" (Ironically, many of the people currently studying artificial neural networks received their Ph.D.s in cognitive psychology.) In point of fact, science is a method rather than a predefined set of topics or body of knowledge, and anything can be studied via the scientific method - yes, even people's thoughts, feelings, and behavior. I've found that people who accuse psychology of lacking rigor tend to misunderstand how it is studied. It doesn't help that psychologists are routinely portrayed by the media as headshrinkers who analyze your dreams while you lie facing away from them on a couch doing free associations . Suffice it to say that this is not an accurate view of what psychology is all about. (I wonder if wearing white lab coats and carrying around colorful beakers could improve our reputation? ) Certainly psychology faces a unique set of challenges by studying what is probably the most complex and dynamic thing on the planet, namely the inner life of human beings. I could rather easily make the case that studying things as relatively simple as elements and atoms is "science for pussies" (but I won't ). Rather than becoming dismayed at the difficulty of scientifically studying people's thoughts, feelings, and behavior, I say that we delve in and try to sort things out as best we can. Psychology is a young science and there is so much left to learn - I find it to be very exciting.

[Sornaensis]

Have Neural Networks ever been observed to become cognitive in any way?

[Xei]

Yeah, it's called a brain. What exactly is it about psychology that strikes you as being, ahem, "for pussies?" (Ironically, many of the people currently studying artificial neural networks received their Ph.D.s in cognitive psychology.) Everybody knows it isn't a 'proper' science. There's so many unfalsifiable or even meaningless models... there's just no way you can find out how the brain works via studying a tiny subset of its actions and reactions, and in an unthorough way. And if I could point out; none of these people have so far made any progress in the field. The only way to do it is to do it properly. It's exactly the same problem that was faced with the 'what is a gene and how does a gene work' issue. There's no way you can work it out by studying entire organisms. To explain it properly, you have to work out what the actual code is, and how it works. That has now been achieved, and the benefits to science have been huge. Did you say you were a psychologist, by the way? If so I'm not trying to be rude... just honest. And if I'm just being ignorant perhaps you could clear some stuff up.

[Sornaensis]

Whoops, I meant Artificial Neural Networks...

[The Cusp]

'Cognitive psychology' or whatever are never going to get us anywhere. The only way forwards is a brute force cracking of how the brain actually works. That's exactly what artificial neural networks allow us to do.

[Xei]

Yeah, I know. Although they still haven't actually figured out how they work yet. Whoops, I meant Artificial Neural Networks... No, not yet. In the same way it was many years before uranium 235 was observed to undergo stable fission. They're working on it.

[The Cusp]

Have Neural Networks ever been observed to become cognitive in any way? The short answer is no. But... There are a lot of experiments where it kind of, sort of becomes cognitive in some way. All very technical, but that's where it get interesting and raises all sorts of philosophical questions. Plus there's all kinds of gray in between ground. Stuff like biological computers, mice with artificial neural nets connected to their brains. Never anything concrete, but the potential is always there, teasing.

[Sornaensis]

You mean like the Jabberwacky program that can link words and emotions together and respond semi intelligently? Or something different? Neural Nets are so abstract...