Slashdot Mirror
Silicon Brains That Think As Fast As a Fly Can Smell
Nerval's Lobster writes "Researchers in Germany have discovered what they say is a way to get computers to do more than execute all the steps of a problem-solving calculation as fast as possible – by getting them to imitate the human brain's habit of finding shortcuts to the right answer. A team of scientists from Freie Universität Berlin, the Bernstein Center Berlin, and Heidelberg University have refined the idea of parallel computing into one they describe as neuromorphic computing. In their design, a whole series of processors designed as silicon neurons rather than ordinary CPUs are linked together in a network similar to the highly interconnected mesh that links nerve cells in the human brain. Problems fed into the neuro mesh are broken up and processed in parallel, but not always using the same process. The method by which neuromorphic processors handle problems varies with the way they're linked together, as is the case with neurons in the brain. The chips are designed to copy the layout and functions of brain cells, but the way they're interconnected is based on another highly efficient biological model. 'The design of the network architecture has been inspired by the odor-processing nervous system of insects,' said one of the researchers. 'This system is optimized by nature for a highly parallel processing of the complex chemical world.' In tests using real-world datasets, the prototype was able to match the performance of specialized Bayeseian pattern-matching systems. Even better, the stable decisions reached by 'output neuron populations' take approximately 100 milliseconds, which is the same speed required by the insect nervous systems on which the network design is based, according to the paper."
Solutions that evolution produces (whether real or simulated) typically suck, as they are typically just good enough for the training criteria and may even completely fail longer term. This really is nonsense, unless you have very low quality requirements. And, unlike a solution based on understanding how to solve something, this bio-inspired stuff cannot easily be improved incrementally from seeing how it performs in practice.
Most ACs are not even worth the keystrokes to insult them. Be generically insulted by this and ignored otherwise.
You're pointing to articles on high end mammalian brain structures when TFA is referring to the most basic structures in an insect brain. Also, this blanket assumption that no one could possibly understand the complexity of a small group of neurons is way out of date.
Olfactory circuits are pretty well understood. This isn't the first simulation of neurons mimicking an olfactory bulb at the single neuron level. We've been watching videos and seeing presentations of these models for years now. What is neat, here, is that they're modeling a somewhat realistic hardware instantiation of a model (as in, this is something which maybe could be built).
I come at this from the other end. I make the chemical sensing hardware that mimics the response of a biological chemical sensor (an artificial insect 'nose'). There are long running collaborations between my field and neuromorphic computing folks to develop a combined sensor-processor that can electronically understand smell in the same way a living thing does. I have to sit through their talks on modeling neurons, and they have to sit through my talks on nanosensor arrays.