From: Rüdiger Koch (rkoch@rkoch.org)
Date: Sat Apr 06 2002 - 08:01:14 MST
On Saturday 06 April 2002 16:30, Eugen Leitl wrote:
> On Sat, 6 Apr 2002, Rüdiger Koch wrote:
> > ASCI White is most likely sufficient to emulate a Fruit Fly. Blue Gene
> > (due 2004) should be sufficient to emulate a mouse. They are general
> > purpose
>
> You're confusing computational equivalent and biologically realistic. A
> 100 kNode Blue Gene filled with the right code and data might be capable
> of mighty deeds (we don't know exactly how mighty), but this doesn't
> happen as long as we don't have that magic conglomerate of code and data,
this is the point, of course. Without a ***close enough*** approximation of
the computation every neuron does and the topology of them, Blue Gene doesn't
help. But you were talking about the hardware!
> > computers. If the same effort would be put into a neuro machine then the
> > hardware side of emulating a human brain is likely in reach now. I am
> > saying most likely because we still don't know if neurons are way more
> > complex as we think they are (e.g. Penrose).
>
> Penrose is smoking heavy crack here, but biological neurons are not the
> caricatures you deem them to be.
What exactly do you mean by "biological realistic". To me it's realistic if
the resulting behavior of the ANN cannot be distinguished from the behaviour
of the original brain. There are gray areas in that definition, e.g. do you
need to model the influence of drugs. I'd say this is irrelevant. Virtual
life might get boring without virtual crack, though ;)
For the nematode - a neuron has less than 20 dendritic neurons and less than
20 axonal connections. The output can be modeled as a single real number.
Timing seems irrelevant. So what kind of expensive computation of F would
give me Axon=F(preNeuron_i)? The are gap junctions. I don't know whether they
need to be included in the model or not. They probalby should. But I am
really lacking the imagination of what you intend to compute with less than
6000 synapses and maybe 400 gap junctions which would require 20,000..1M
processors for a real time sim. You give every worm neuron up to 3
TeraOps/s.
The true bottleneck is not the hardware power of IBMs high end, it's scanning
and observation technology.
-- Rüdiger Koch http://rkoch.org Mobile: +49-179-1101561
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:13:16 MST