Emulating human brain

Discussion in 'Intelligence & Machines' started by vpidaparthi, Dec 30, 2003.

Thread Status:
Not open for further replies.
  1. vpidaparthi Registered Member

    Messages:
    3
    How many processors and how much memory does it take to simulate a human brain? Isn't it a good idea to simulate a human brain and store all the knowledge in it and let it gain experience with time? This artificial human brain can live forever and gain experience.
     
  2. Google AdSense Guest Advertisement



    to hide all adverts.
  3. sargentlard Save the whales motherfucker Valued Senior Member

    Messages:
    6,698
    Hmm you're thinking in terms of current PC specs? Not particularly a good idea. To emulate a human brain you need a analog computer....a self checking and correcting program...unlike todays PCs which alert the user of the problem but fail to correct it on their own and make preventive measures so such a problem does not occur in the future (No certain features in XP don't count as a step toward the human mind, infact windows is a huge step back.)

    I don't know if you could put an exact amount on how much processing power is needed because certain tasks that we do as humans which are quite easy to us require near infinite processing power yet a single 2.0 ghz processer can do raw calculations at blinding speed which a human brain can't.

    What I imagine is quantum computing to properly emulate a human mind, it is still in early stages (Last i heard IBM was in the lead of quantum computing research)....Moore's law will eventually be broken.
     
  4. Google AdSense Guest Advertisement



    to hide all adverts.
  5. hlreed Registered Senior Member

    Messages:
    245
    Quite simple.
    Count and name each neuron.
    Replace each neuron with a computer that does
    Axon = N(synapses) where N is a neuron function.

    This is dynamic. There is no place to store data except in the neurons. That is what N does. N is simple. It is the combinations that get you.

    A brain is nothing like a computer.
     
  6. Google AdSense Guest Advertisement



    to hide all adverts.
  7. AntonK Technomage Registered Senior Member

    Messages:
    1,083
    hlreed,

    When do we get to see something you've created. You've been posting here for a LONG time about all your research, and you sell a book, but show us something you've done.

    -AntonK
     
  8. kmguru Staff Member

    Messages:
    11,757
    Another way may be emulating the DNA expression using cellular automata math. We may not get a talking human, but we might get a music crunching savant.

    Please Register or Log in to view the hidden image!



    The answer is in the blue print called DNA. These days, we simulate nuclear explosions and automobile designs through simulation....
     
  9. Stryder Keeper of "good" ideas. Valued Senior Member

    Messages:
    13,105
    It seems the current trend of thought is to take a Neuron and place it into a mechanised form, personally I think that is where the generation of a Brain Emulation system would fail as your trying to replicate something piece by piece without fully understanding the overall functions of the real brain structure.

    A preportion of learning how to develop a true Brain Emulation system would be to place a human mind into the range of passive scan equipment, and to generate a 3D virtual representation of the human mind.

    To generate an emulated Consciousness
    While the person is in range of the equipment you would have to question them about past memories to try and fit together the overall pieces that make up the psychological past of the brains functions (although you would not have an accurate record of how the brains functions, cells etc evolved up to the point they enter the passive scanning range)

    The understanding here is you would then have to take all the overall patterns that are created (Vectors of thought and cellular changes over time etc) and place them back together to generate as seemlessly as possible a conscious structure.

    To Emulate the brain
    This is the main reason for having to emulate the consciousness, as without a system understanding what interactions to generate, what areas of the brain to stir etc for particular outcomes, how could you actually emulate a true brain?

    So conclusively if you wanted to emulate a human brain, it's whole existance would have to be placed within a large network of computers that parallel process the countless passive sensor logs to generate overall thought processive functions.

    There is then the potential to use active sensors to generate doppler effects of thought functions to stimulate patterns of thought to which in turn generate memories or even place complete structures into the mind in the form of a dream or imagination.

    (There is then the possibility of harnessing such illustrations with a method of allowing the conscious brain to decide how the illustration should be master, as if they have the ability to create a "directors cut" which in turn allows an individual to create animations or build elaborate film sequences etc.)
     
  10. artfldgr Registered Member

    Messages:
    5
    each neuron can have well over 100,000 connections and there are over a billion of them. it was recently shown that glial cells are important as well so up the ante there. it will be a long long long time before we could muster just equivalent computing power, let alone technique. using the methods that we use to make chips that would be one hot puppy. our system communicates using cold methods, or reasonably cold methods. each cell generates its own energy from chemical methods (ATP), and so there is no need for a centralized power source to feed the whole system. the sugars and fats, and oxygen is a cold power system as opposed to biological system..
    the other problem is that our computing systems are time ticked. they all step forward based on a clock tick. while there ARE chips that dont do this, until we understand how to create self adapting archetectures with compartmentalizaition we will have other problems...

    the problems related to this issue are fairly insurmountable with our technology and up and coming technology.. at least in our lifetimes. this is why there are chips that use groups of living cells that are being tested.. i would not look forward to seeing anything like this for a very long time, if at all (it might not be needed given that what we want in machine intelligence is not new complete thinkers but compartmentalized thinking sub functions).

    as far as storing thoughts for recovery.. well.. first you would have to make a machine that can read the mind and its state completely.. hmmmm i can see the public smashing that one.. and then you would have to figure out a way to impose a state onto a brain.. hmmm there is another machine that would get smashed..

    why for both.. well one would leave no secrets and would constrain how we think, we would be afraid to thing for fear that an imployer would read it all and find us unsuitable because we have a different or novel idea..
    the other would lead to one group in power using the machine to change those that they thought werent thinking the way they should.

    all of this ignores abnormal structures, and variance in chemical operation based on activiated and quiessence genes..

    too much..

    artfldgr..
     
  11. kmguru Staff Member

    Messages:
    11,757
    The biological system also works on time ticking too. There are millions of chemical integrators that calculate output based on feedback signal. They all have very fine tuned integrators and proportional controllers which is impossible to co-ordinate with out some type of time reference. Just because the doctors have not found a single clock cell does not mean the mechanism is not there.

    As to the mind read, I am sure there is a central command for a higher level function (not the distributed body function). If we can locate that - it should not be very difficult to plug in an I/O units to upload/download the data.

    Just a thought...
     
  12. androgen Registered Senior Member

    Messages:
    44
    i think a wild-ass-guess estimate should go like this. take the number of synapses in the brain, divide by the ratio of switching speed of silicon versus neurons, and that should be the number of gates on a chip.

    the architecture needed to implement all this mess would probably be nothing like anything we know today though, it would certainly be massively parallel and the number of gates we're looking at here is also going to be massive.

    i dont think it really matters whether we go digital or analog.
     
  13. kaduseus melencolia I Registered Senior Member

    Messages:
    213
    Some people still don't get that you need a brain to be connected to senses.
    The senses aren't static so you need muscles.
    Now you need a body to mount everything.

    Anyway, you won't get it done using windows.
    You won't do it using parallel connections either.

    Lets say you could manage 1,000 neurons per 'computer'.
    1,000,000,000 neurons = 1,000,000 'computers'.

    we could put the 'computer' on a small board and layer the boards into blocks.
    10 boards per block = 100,000 blocks
    46x46x46 blocks = 97,336 blocks

    10cm or 4" per block = 4.6mx4.6mx4.6m or 15'x15'x15'

    Thats a big expensive machine!

    How many neurons in a cat?
     
  14. kmguru Staff Member

    Messages:
    11,757
    Lets say we reduce the 15'X15'X15' to 1.5"X1.5"X1.5" (like we did from vacuum tube to ICs) and the price too....

    Thats a small cheap machine!

    Please Register or Log in to view the hidden image!

     
  15. androgen Registered Senior Member

    Messages:
    44
    kaduseus i call bullshit. neuron is big and slow, transistor is small and fast. we should be able to emulate the human brain in something which was at least as small.

    as i see it the real problem with processors is they are manufactured in pretty-much 2D while the brain is 3D.

    a typical processor die is one layer of transistors 1 cm by 1 cm in size. if we want to increase transistor count by factor of 10,000 we would have to bump it to 100cm by 100cm. this would create ridiculously long paths that would take a long time to conduct a signal and would generate a lot of heat conducting it.

    but what if instead we pile 1000 circular 2" dyes one on top of each other, we would still have the same transistor count but in a compact package without the problems mentioned above.

    the only problem like i said would be cooling. how do you cool a processor that makes say 10 kilowatts heat ? i will tell you. we could make our disk porous ( with nice straight channels ) and put ice-cold water through it under pressure.

    we would need to come up with some new technologies to make this workable but this is the next logical thing to do imho ...
     
    Last edited: Apr 9, 2004
  16. Dapthar Gone for Good. Registered Senior Member

    Messages:
    203
    I'm surprised that everyone is so focused on the hardware aspect of the problem. Given time, (say, within 50 years maximum), there will be more than enough computational power to simulate the brain. However, to put it bluntly, hardware has never really been the problem. Lazy computer scientists simply use it as a crutch to justify their lack of feasible AI code.

    For example, if there was a solid idea of how to simulate a human brain, it could be written now, and run on current machines. It may cause even the best super computers to plod along while running the code, but it would still run. Also, it's not like the routines that are constructed heavily rely upon new algorithmic techniques.

    For example, consider the field of computer vision, which, at it's most fundamental level, has the same problems to solve as those faced by true AI researchers. Facial recognition software (see Eigenface) uses 170 year old Mathematical techniques (eigenvalues), to solve the problem of recognizing distinct features of faces. Moreover, the calculation of this data is essentially the extent of the algorithm! No "new computer science paradigms" were required.

    Prior to this, I recall the cries of critics in the field, stating that facial recognition software wouldn't be feasible until computers sped up considerably, due to the predicted amount of data necessary to uniquely identify a face. Granted, this software is still in its infancy, and not without its problems, but even in its early state, it can analyze hundred of individuals per second.

    To conclude, I would just like to state that the hardware will eventually have the capacity to simulate the brain. However, I doubt that we will see a true simulated brain for a while, unless some of these AI researchers get their acts together, and stop depending on hardware to fix their problems.
     
  17. Stryder Keeper of "good" ideas. Valued Senior Member

    Messages:
    13,105
    Well thats pretty much the realm of nanotechnology would come into play since you would have to develop singular nodes that can computer simplistic values that can parallel process as a whole system.

    Also the programming of an AI system currently utilises too much "Meta" information as aposed to solid substance, namely an AI program is suppose to understand "four apples", but neither understand what "four" or "apples" actually are physically.

    This would mean that a full synthesis of how we spectrally analyse information would have to come into being, The ability to cover all five senses that we take for granted and the sixth of Tangible thought. (the method of converging the matrix mathematics of all senses together)

    You could further disect the problem by introducing how a system would cycle information for memory, and how to "Cluster" memorised thought.

    Although if you were to look at the basis of the human brain, you would have to realise that there is a blue print pattern of RNA, that utilises building blocks to define the parameters of the brain, but the neurological pathways themselves don't follow that rule as they have to "Develop themselves" based upon the sensory interactions.
     
Thread Status:
Not open for further replies.

Share This Page