Posted by:
Henry Bemis
(
)
Date: August 15, 2013 08:46PM
"True North is built on a network of “neurosynaptic cores” that place memory, processing, and communication close to one another so they can operate in parallel, much as they do in the brain."
"Just as von Neumann computing required a specialized programming language (Fortran), Modha says SyNAPSE’s new architecture needed the same—so his researchers wrote a new language from the ground up, and began organizing and simplifying it for future scientists and software developers."
"Early on, programmers had to code individual neurosynaptic cores, but the language now includes 150 “corelets,” or groups of cores with similar functionality (eg., sound perception, edge detection, or color identification). Developers only need to know the general function of a corelet to integrate it into an application."
I noted the implied claim that the model in question involved something more than a standard "Von Neumann" computer. But it appeared that the distinction was only that of parallelism, and not truely connectionist. In short, we still have a Von Neumann computer modeling a connectionist architecture. At least, that is what it appears. What raises my doubts is that the brain is not digital, thereas computers are.
Here is a quote from connectionist, and AI advocate, Patricia Churchland:
"Pushing at the bounds of the simulation paradigm are ambitions to go beyond simulation of a neural circuit to construction of a synthetic neural circuit--for example in a chip. Applying this to a specific instance, we are referring to the difference between contructing a device that can respond to real, as opposed to simulated, light; . . . The contrast we have in mind between a simulated retina and a synthetic retina is roughly this: in a simulated retina the input will be numerical values corresponding to properties of light such as wavelength and intensity, and the output will be numerical values corresponding to information carried by the ganglion cells. In a synthetic retina, the input is real light, and the output consists of electrical signals carried not in spike trains but in current pulses. Following this line, the constructive problem concerns how to go from simulating in a computer to making synthetic retinas, nuclei, spinal cords, cortices--in short, how to make synthetic brains."
(Churchland and Sejnowski, THe Computational Brain.)
What we have here is apparently still a classical computer simulation. But I could be wrong about this.