5/4/07

First they came for the mice


In a thread I can no longer find over at eripsa, I argued that it is foolish to think that we won't soon have true artificial intelligence because soon we'll be able to simulate all of the activity of the human brain, albeit a little slowly.

Now I see via egg that we are one step closer to this nightmare:
Cory Doctorow: IBM researchers have modelled a mouse's brain at 10 percent speed -- and what can be done at 10 percent speed today can be done at 1000 percent in a couple cycles of Moore's Law. Super-intelligent virtual mice ahoy!
Neurobiologically realistic, large-scale cortical and sub-cortical simulations are bound to play a key role in computational neuroscience and its applications to cognitive computing. One hemisphere of the mouse cortex has roughly 8,000,000 neurons and 8,000 synapses per neuron. Modeling at this scale imposes tremendous constraints on computation, communication, and memory capacity of any computing platform.

We have designed and implemented a massively parallel cortical simulator with (a) phenomenological spiking neuron models; (b) spike-timing dependent plasticity; and (c) axonal delays.

We deployed the simulator on a 4096-processor BlueGene/L supercomputer with 256 MB per CPU. We were able to represent 8,000,000 neurons (80% excitatory) and 6,300 synapses per neuron in the 1 TB main memory of the system. Using a synthetic pattern of neuronal interconnections, at a 1 ms resolution and an average firing rate of 1 Hz, we were able to run 1s of model time in 10s of real time!

Gonna need a lot of virtual cheese.

No comments:

Post a Comment

eXTReMe Tracker