jjj (jaceklk at hotmail.com) wrote:
: Hi all,
:: I just read an article in Science magazine about that monkey having a scan
: of her brain with fNMR (dynamic, real time magnetic resonance) and I had
: that thought: I think that researchers struggle to create a human-like,
: artificial brain. They try to teach that brain on the input from environment
: only, but how about if you can have a human fNMR scan with all different
: thoughts as an additional input. You can create easily a vast amount of data
: for a artificial neural net!
: Another words, you would input both visual/auditory input (same as that real
: brain receives) AND the blood flow in biological neural network (fNMR scan).
: Goal wold be to recreate a net which would behave the same way as the real
: Advantage would be that you can easily test the new artificial brain if both
: the fNMR of real brain and the activity in artificial neural net are the
: same. Also, I'm not sure but I feel that additional,easily acquired info,
: would help in creating the net since it give more information about the
: hidden nodes/layers of the neural net.
: What do you think? Would that help a little in ultimate goal? :)
: (I realize that the real NN are very complex and that simulating that vast
: number of complex neurons is not possible today.)
The first thing you would need to do is to PROVE that the brain
works in the same manner as a neural net. That should keep you
busy for a while.
Hint: There's a lot of things going on in neurons that no neural
net that I know of can mimic. This fall, I'm going to take a
rather intense course in neural networks, but without the delusion
that I'm studying the brain in the process.