Brainstorm

Canon PowerShot SD700 IS

ƒ/2.8

5.8 mm

1/60

3D computer models of neural growth and patterns of firing (modeling a cortical column for synthetic olfaction). Implemented on $2K of NVIDIA GPUs from Fry’s Electronics — a DIY supercomputer.

Left Brain: dendrites grow along chemical gradients, densely filling all available space, and exploring approximately 500 possible connections for each ultimate synapse formation.

Right Brain: neural model with detailed modeling of the computation that occurs in the dendrites. Earlier, simplistic neural network models assumed that the cell body was the aggregating locus of computation.

9 responses to “Brainstorm”

angry shock

Apr 6, 2007 at 7:29 pm

a brain running on windows oh my god! 😉

Reply
Today is a good day

Apr 8, 2007 at 5:15 pm

Cool graphics!

I used to work in the fiield of Neural Networks – I wish I had access to the computing power people have today!

—
Seen on your photo stream. (?)

Reply
HelpFeedAlex

Apr 10, 2007 at 4:45 am

Neat! Just steps away from being able to record people’s dreams in hi-def!

Reply
juandesant

Apr 13, 2007 at 3:31 pm

Nice modelling tool! 😉

As it happens, I’m an admin for a group called Computer Simulations, and we would all love to have your photo added to the group, as I think it is a perfect match.

Of course, if you have computer snapshots of the images in the monitors, all the better!

Reply
quickest part

Sep 30, 2008 at 4:18 pm

Looks like two 23" cinema displays. What are they hooked up to?

Reply
jurvetson

May 19, 2009 at 11:16 pm

Rows of NVIDIA chips. DIY supercomputer

Reply
Amit909

Mar 13, 2011 at 6:31 am

nice

Reply
jurvetson

Jun 22, 2013 at 5:22 am

the math behind the model:

The voltage spike will propagate, with gain, along the dendrite, from one finite element region to the neighbor, not as a free flow of electrons, but as a bucket brigade of opening and closing ion channels, like fingers cascading down a very long flute.

The dendrite with myriad branches is a fundamental locus of computation, not just the neuron cell body.

Reply
jurvetson

Dec 20, 2016 at 5:13 pm

Now that GPUs have revolutionized the efficacy of neural networks, Evolved Machines CEO Paul Rhodes reflects: "The GPU impact on compute is beautiful. We had CUDA 0.1, and one of the first cards (it was called the G-80), handed to us in a bag, pre commercial availability. I will never forget the 80x speedup we got on our neuron simulations when we ported to that platform. It is so native for neural computation."

Reply

Brainstorm

9 responses to “Brainstorm”

Leave a Reply Cancel reply