imported_IIB
Member
bump
-
We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.
Beyond AI Today:
Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Matthias99
Diamond Member
First, you suck at teh quoting. I'm fixing this one, but please try not to mangle your posts so badly that they're unreadable.
Again: please come back when you have a clue. Saying they are 'more advanced in cognitive capabilities' is meaningless drivel.
Again, you're just blathering on randomly.
Yes, they're adaptable.
Haven't read the book. Certainly the theories being discussed there are not universally accepted.
I didn't say any of those things.
No, just like an individual neuron is not 'intelligent' either, no matter how many interconnections it has.
You'd have to understand them more completely first to model them.
Where the hell did this one come from?
Uh... what?
That would be why they also model the interconnections between neurons.
It's taken explicitly into account in some of them.
How many computers can you put in a network?
This question basically implies that you don't really know much if anything about signal processing or computational theory.
Again, you're assuming that the neural medium has something 'special' about it that is impossible to capture or model in a useful way. This kind of thinking is really not shared by most of the scientific community.
...that wasn't really a question. And yes, it is a "step", a VERY early one.
And you're considering them a magical black box that can never be understood. "simple" was perhaps an understatement of the problem, but the complexity here is limited in scope.
That's not really what Occam's Razor is. 😕 And no, this is not 'obvious' to a lot of people who actually do this research, and they generally know more than you.
Again, you're theorizing that any of this will exist anytime soon, and that it will all work out nicely the way you think it should.
It might be, it might not be. Research really isn't far enough along yet to say. Using living cells has some inherent disadvantages -- they're fragile, they're harder to build exactly the way you want them, and trying to interface between analog and digital systems of any substantial size is a hard problem in itself.
...Uh, yeah. You get right on that.
Earth to poster, Earth to poster, come in please.
2001-2003. I've followed some of the research from that lab and others since then.
Are they going to keep working on this? Yes. Is it at the point where we can throw out our PCs and build everything using neurons? No. Will we someday get to that point? Maybe, but I think not for general-purpose computing.
What am I not "seeing"? Everything you're talking about is still an idea or on the drawing board somewhere.
Define 'high-end stuff'. The lab I worked in was doing research in intracellular recording and modelling, partially under a DARPA contract. Since I left they've been doing some human trials (they run experiments on patients already undergoing neurosurgery, with their permission).
You're basing all of this on the thinking that DARPA is some shadowy super-organization that has unlimited resources. They are not. They are, for the most part, an organization that guides, funds, and organizes research.
Again with the conspiracy theories.
I tried. Yours doesn't add up.
Again: please come back when you have a clue. Saying they are 'more advanced in cognitive capabilities' is meaningless drivel.
Again, you're just blathering on randomly.
Yes, they're adaptable.
Haven't read the book. Certainly the theories being discussed there are not universally accepted.
I didn't say any of those things.
No, just like an individual neuron is not 'intelligent' either, no matter how many interconnections it has.
You'd have to understand them more completely first to model them.
Where the hell did this one come from?
Uh... what?
That would be why they also model the interconnections between neurons.
It's taken explicitly into account in some of them.
How many computers can you put in a network?
This question basically implies that you don't really know much if anything about signal processing or computational theory.
Again, you're assuming that the neural medium has something 'special' about it that is impossible to capture or model in a useful way. This kind of thinking is really not shared by most of the scientific community.
...that wasn't really a question. And yes, it is a "step", a VERY early one.
And you're considering them a magical black box that can never be understood. "simple" was perhaps an understatement of the problem, but the complexity here is limited in scope.
That's not really what Occam's Razor is. 😕 And no, this is not 'obvious' to a lot of people who actually do this research, and they generally know more than you.
Again, you're theorizing that any of this will exist anytime soon, and that it will all work out nicely the way you think it should.
It might be, it might not be. Research really isn't far enough along yet to say. Using living cells has some inherent disadvantages -- they're fragile, they're harder to build exactly the way you want them, and trying to interface between analog and digital systems of any substantial size is a hard problem in itself.
...Uh, yeah. You get right on that.
Earth to poster, Earth to poster, come in please.
2001-2003. I've followed some of the research from that lab and others since then.
Are they going to keep working on this? Yes. Is it at the point where we can throw out our PCs and build everything using neurons? No. Will we someday get to that point? Maybe, but I think not for general-purpose computing.
What am I not "seeing"? Everything you're talking about is still an idea or on the drawing board somewhere.
Define 'high-end stuff'. The lab I worked in was doing research in intracellular recording and modelling, partially under a DARPA contract. Since I left they've been doing some human trials (they run experiments on patients already undergoing neurosurgery, with their permission).
You're basing all of this on the thinking that DARPA is some shadowy super-organization that has unlimited resources. They are not. They are, for the most part, an organization that guides, funds, and organizes research.
Again with the conspiracy theories.
I tried. Yours doesn't add up.
imported_IIB
Member
No it's not. I hate to break it to you but Cognitive 'computing' is more than just number crunching.
Again, you're just blathering on randomly.[/quote]
You're dismissing neurons as if theyre logic gates, still.
Yes, they're adaptable.[/quote]
Haven't read the book. Certainly the theories being discussed there are not universally accepted.[/quote]
And neither is your THEORY of "they're not really all that powerful (or unique) in terms of computational ability". What books have you written? I'm citing experts here, you're just dismissing things with your opinion.
I didn't say any of those things.[/quote]
"they're not really all that powerful (or unique) in terms of computational ability" I'm pretty sure you said 'isnt feasible" in her esomewhere. Ok, so then if my theory isnt right, and conventional methods arent the key, then what is? Note that in a couple places I refered to quantum computing as conventional because it's still number crunching. Those photons dont exactly store actual memories.
No, just like an individual neuron is not 'intelligent' either, no matter how many interconnections it has.[/quote]
Explain this finding by Christof Koch:
http://cbcl.mit.edu/cbcl/news/files/kreiman-hogan-5-05.htm
You'd have to understand them more completely first to model them.[/quote]
Of course, along ith all of these other things I'm mentioning and the ones that I havent yet.
Where the hell did this one come from?[/quote]
THis is an important dynamic when determining the feasibility of neuron processors, they self repair and can potentially reach lifespans of over 100 years. How many 100 year old complex silicon chips or hard drives are you aware of?
Uh... what?[/quote]
This is an important dynamic in judging if we're really going to build a mathemattical model of neurons, to achieve cognitive computng.
That would be why they also model the interconnections between neurons.[/quote]
Even if they completly modeled rat cortical neurons that's a long shot from modeling far more powerful human neurons typeS. But considering all of the dynamics of neurons (that we STILL dont fully understand) that's still a long shot. Modeling the way they connect is only the tip of the iceberg, its a long way off from FULLY understanding what's realy going on inside them and how to easily (low resource overhead) incorporate their memory abilities into the math model. How is it more feasible to wait possibly several more decades to finally gain a complete understanding of neurons to get a proper math model when we can possibly skip that and build brains our selves?
It's taken explicitly into account in some of them.[/quote]
That's fine, but its seriously another example of the type of hardware overhead this software would demand.
How many computers can you put in a network?[/quote]
Your response is out of context. The point is we could A: potentially hook several massive brain arrays up to single quantum rigs, OR B: hook up massvie arrays or quantum rigs including massive storage arrays and gross amounts of RAM.
It's a matter of perfecting the brains, that couldpotentially live for over 100 years once grown and taught. Less conventional hardware components would have to be used, and replaced every so often. Instead of replacing and upgrading a countless number of CPU's, RAM and permanent storage they would only have to keep the brains stable and replace far less conventional components.
This question basically implies that you don't really know much if anything about signal processing or computational theory.[/quote]
It could have been worded better. The point is that neurons are analog, conventional electronics are binary. It seems like you're talking about running mathematical models (THAT COMPLETELY REPLICATE NEURONS IN ALL OF THEIR COMPLEXITY) over conventional computers. If you were smart you could have brought in the emreging quantum computers capabilities, but i would stiill have my argument that complete math models to completely replicate neuron power is a LONG way off.
Again, you're assuming that the neural medium has something 'special' about it that is impossible to capture or model in a useful way. This kind of thinking is really not shared by most of the scientific community.[/quote]
Again, you're assuming that neurons are just simple logic gates. I never said that biologically inspired neural nets cant be powerful, but I think your assumption that neurons are essentially logic gates is completely unfounded.
...that wasn't really a question. And yes, it is a "step", a VERY early one.[/quote]
lol. Ya I forgot to finish that one. IT should have said "Does this realy appear to be a critical step in surpassing human cognition? But you said it: "a VERY early one". Exactly my point. You pretend as if conventional AI is the asnwer, blasting my theory as being too far off, but SO IS your conventional thinking. Which could come first, or more importantly get us there first is the key. Since all you're personally seen is simple neuron nets you assume that there's no way neuron brains is possible.
And you're considering them a magical black box that can never be understood. "simple" was perhaps an understatement of the problem, but the complexity here is limited in scope.[/quote]
But their still not understood and everyone assumes that we'll completely model them before we can just put them to use?
That's not really what Occam's Razor is. 😕 And no, this is not 'obvious' to a lot of people who actually do this research, and they generally know more than you.[/quote]
Well I thought occams razor is:
Given two equally predictive theories, choose the simpler, and The simplest answer is usually the correct answer.
The point is that you havent acually demonstrated that their not, you've just been dismissing the potential of live neurons. We have live neuron power at our dispossal, but we dont even fully know how powerful they are but we can tell theyre more than simple math gates. We're trying to study them to achieve cognitive computers, but we're only studying simple rat cortex neurons - and you think that that can get us there first?
Again, you're theorizing that any of this will exist anytime soon, and that it will all work out nicely the way you think it should.[/quote]
How can say for sure. Have you been analyzing all of these finds and all of the data scientific data on the web and in the TeraGrid? I'm sure they have been. Those contrats you talk about arent just for the hell of it, they're for the obvious:
http://www.darpa.mil/ipto/programs/bica/index.htm
It might be, it might not be. Research really isn't far enough along yet to say. Using living cells has some inherent disadvantages -- they're fragile, they're harder to build exactly the way you want them, and trying to interface between analog and digital systems of any substantial size is a hard problem in itself.
Exactly: "It might be, it might not be." So dont go dismissing this just yet. That answer also applies to how far along they could be, right now. We cant know for sure, so therrefor it'd be foolish and ignorat to proclaim that there's no way its not going to be used. It could be next year, it could be by 2010. Honestly, it could already be old news.
...Uh, yeah. You get right on that.[/quote]
This isnt my goal, I'm just trying to determine what they can, will and are doing.
Earth to poster, Earth to poster, come in please.
I'm not the mad scientist here, it's really you people. I didnt think this stuff up, it was people like thos einvolved in this list of research that I'm citing, and others like you who are involved in these sciences, and dont try telling me that it will all always be used for good things like treating parkinsons or alzheimers.
2001-2003. I've followed some of the research from that lab and others since then.[/quote]
Again you're basing your arguments from the simple low level studies that they're doing. It isnt just 'your' lab, it's virtually all US labs doing their share of DARPA's research. They sit at the top fitting all of the pieces together, and for what? BICA
Are they going to keep working on this? Yes. Is it at the point where we can throw out our PCs and build everything using neurons? No. Will we someday get to that point? Maybe, but I think not for general-purpose computing.
I never said that "we" would be throwing out our old computers. The government will not allow us to have cognitive computers, probably not for at least a decade until after they've done it long after they have far superior ones. By then they plan on having us all neurally jacked into the system, we wont need them anyways.
What am I not "seeing"? Everything you're talking about is still an idea or on the drawing board somewhere.[/quote]
And all of your arguments are based on assuming tha tneurons are essentially simple, and that there's no way that the government isnt doing this because you've seen lab cultures of up to about 25K neurons (it's beyond "dozens" as you misleadingly described it earlier). It sure is on the drawing board, has been for soem time.
Define 'high-end stuff'. The lab I worked in was doing research in intracellular recording and modelling, partially under a DARPA contract. Since I left they've been doing some human trials (they run experiments on patients already undergoing neurosurgery, with their permission).
You're basing all of this on the thinking that DARPA is some shadowy super-organization that has unlimited resources. They are not. They are, for the most part, an organization that guides, funds, and organizes research.[/quote]
Not entirely. i forgot to apply the question mark afte rthe second question. But ya, it's important to note who we're talking about here: the top (known) military technology agency in the US, which has the top technologies in the world.
Again with the conspiracy theories.[/quote]
Again with assuming that conspiricie sdont exist and believing that the top military agency doesnt have secrets. :lol:
I tried. Yours doesn't add up.[/quote]
How doesnt it? There's enough material on the table, explain how it's impossible and absolutely infeasible, dont just nitpick each point in tis ever growing debate. Explain how you know for a fact that neurons are essentially simple, and how it's not possbile to do what i'm saying, and how DARPA has absolutely no secrets. Feel free to cite some actual science instead of just dismiss all of this for the mathematical model that the conventionalist still think is the answer.
VirtualLarry
No Lifer
Sounds like.. The METACORTEX! 🙂
For some additional ideas and information, check out a game published a few years back for Windows PC's, called Galapagos (sp?). In it, you controlled the environment for this spider-like creature, and it learned from its environment. It used some technology called "NERMS".. some sort of self-converging neural learning system.
I have a few theories myself about human minds, cognition, and learning, based mostly on my own observations and half-baked theories. I hypothesize that the human brain actually stores data at a quantum level, like every neuron has some sort of quantum sub-space attached to it, that it uses to store data, and that mostly, there is no real limit to what a brain can store. Also, the way that the mind accesses memory stores, are based off of encoded signals, much like a peer-to-peer network communicates, and "mental associations", are generally based on a shortest-path-first associativity mechanism to a hierarchial echo-like query.
My opinion on "conciousness-altering" drugs, is that some of them cause the operation of the brain to act on a maximal-depth sort of query, in which the first echo responses are delayed, such that a larger number of responses is returned at once to queries, thus the associations made by the mind are much greater in number, resulting in an amplified sense of conciousness.
For some additional ideas and information, check out a game published a few years back for Windows PC's, called Galapagos (sp?). In it, you controlled the environment for this spider-like creature, and it learned from its environment. It used some technology called "NERMS".. some sort of self-converging neural learning system.
I have a few theories myself about human minds, cognition, and learning, based mostly on my own observations and half-baked theories. I hypothesize that the human brain actually stores data at a quantum level, like every neuron has some sort of quantum sub-space attached to it, that it uses to store data, and that mostly, there is no real limit to what a brain can store. Also, the way that the mind accesses memory stores, are based off of encoded signals, much like a peer-to-peer network communicates, and "mental associations", are generally based on a shortest-path-first associativity mechanism to a hierarchial echo-like query.
My opinion on "conciousness-altering" drugs, is that some of them cause the operation of the brain to act on a maximal-depth sort of query, in which the first echo responses are delayed, such that a larger number of responses is returned at once to queries, thus the associations made by the mind are much greater in number, resulting in an amplified sense of conciousness.
VirtualLarry
No Lifer
IMHO, it could theoretically get there, if the system(s) comprising it were designed "better", meaning more resembling the way that the human mind stores and processes information.
I downloaded some papers a few years back from The Santa-Fe Institute, which was a sort of think-tank dealing with complexity/chaos and a number of things, and it actually described the levels of complexity that would need to exist, for that to happen. It touched nicely (in a scientific way), to the concepts explored in the anime "Ghost in the Shell" - aka the "ghost line" - an indication of concious self-awareness of the system. I hope I still have that paper somewhere on my HD.
Learning systems, no matter how they are built, still need a source of input to learn. ("Number 5 is alive! Need input!") The human race could provide a massive amount of input/entropy to learn from. I call that "information inductance".
I think that the reason that early attempts at AI failed, were that they were trying to "build intellegence". I don't think that's possible, it can't be built a-priori, but has to be "grown", from the ground up.
I do think, eventually, that it will happen.
plagiarist
Senior member
So, which motherboard is best for a Rat-brain CPU? Should I order it from Newegg or ZZF?
VirtualLarry
No Lifer
Wait until it is finally discovered, that human brains have built-in "wireless" networking, using quantum sub-spaces, and that humanity can, eventually, become one giant networked conciousness itself, using the very basal fabric of the universe to communicate, conciousnesses linked to other conciousnesses, forming the "metacortex". Allowing the human learning of an individual to be shared with all, the ultimate Hive Mind of Humanity, if you will. Perhaps that is in essence the "Mind of God".
The other interesting thing that I've been exploring recently, is the the possibility that human existance, conciousness, is multi-dimensional - in a way that is greater than our mere four "physical" dimensions. That perhaps, humans in the past, and some "enlightened"/spiritually-inclined humans, managed to awaken their higher-dimensional senses, and that science, which primarily uses "physical" machines to validate their hypothesises, is self-defeating, because those mere machines cannot be used to observe the higher dimensions, only human/living conciousness can. Therefore science will never lead us in the "awakening" process, only truly spiritual religions granting self-awareness can.
I've actually had series of "awakenings" myself, over the last few months, and it has been quite... interesting. Truely mind-blowing stuff, if it turns out to be true. But there isn't really a "scientific" way to prove any of it.
VirtualLarry
No Lifer
That is essentially what the video game "Galapagos" was all about. Now I wish I had played with it a bit more, and not sold it. It looked pretty neat, but I never installed it.
VirtualLarry
No Lifer
That's very interesting. One hypothesis that I had, was that if neurons use quantum sub-spaces to store information, that they might also store what are essentially akin to "stored procedures" in databases.
There are some that believe that the "eyes are a gateway to the soul", and based on my own personal experiences, I hypothesize that humans, actually have a universe unto themselves.. meaning that seeing stars in someone's eyes... might literally be stars, in some sort of nested universe. That universe would supply the necessary entropy, and space for knowledge-storage. I know it's a little far-fetched, but nature has a funny way of being extremely self-similar, at very vastly different levels of scale. I think something along those lines is possible.
If you remember Douglas Adams, the human population and existance on Earth, was really just a grand computational experiment done at the behest of some multi-dimensional beings... but any good computational system is self-virtualizing... leading to nested universes and dimensions of existance. Again, self virtualizing, self-similar, the fractal and chaotic nature of, well, nature.
I still wonder what the Mayans actually figured out about this universe.
If you think about it, though, if humans are five-dimensional (at least!), then their four-dimensional existance is simply a projection (much like a 3-dimensional object casts a two-dimensional shadow on a certain plane), and thus people "teleporting" to different planes of existance, would be willfully chosing their projection into a different four-dimensional universe. (Meaning, I do think that parallel worlds exist, and that we can exist in multiple worlds at once, and that our "soul", exists outside of our four-dimensional space and time. Perhaps that is the "real self", in a higher-dimensional existance, and we (humans), have simply mostly lost our higher-dimensional spaces.)
I have come to a partial understanding that the "Tree of Life", as described in the Bible, may actually be a hierarchial tree of concious existance itself, that conciousness IS life, and that our higher-level existance is not that of seperate animal beings, but rather we exist as branches of a plant-like being. IIRC, eastern religions make reference to Yggdrasil, a "world tree", which seems somewhat of an analogy of the "Tree of Life".
Perhaps God is just a bonsai gardener. 🙂
VirtualLarry
No Lifer
Speaking of which - has everyone seen the anime series "Bubblegum Crisis: Tokyo 2047"? The concept of the "boomers", sub-human animat-like creatures/creations, built from a fusion of human neurons and nano-scale electronics, to build helpers for humanity... and the ensuing ethical debates surrounding their existance. Is DARPA planning on building the same things? Hmm.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 25K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 24K
-
-
