Called neuromporphic chips, the new processors, developed in collaboration between the University of Zurich and ETH Zurich, are about the same size as a neuron, and are capable of performing the same tasks a human brain can accomplish.
One the main hurdles in developing their new processors was configuring the processor network to simulate human neural activity. To do this the team in Zurich built a network of artificial neurons which could be joined together to create a processing module.
In their test, which asked the chip to identify which way a series of bar was moving across a computer screen, the neuromorphic chip’s processing modules successfully identified the bars’ movement. Bully for them, they managed to identify visual input. What’s impressive, however, is that the chip successfully calculated a solution in real-time, demonstrating its ability to merge both visual processing with memory and context dependent decision-making, all of which are believe to be elements of cognition.
"Thanks to our method, neuromorphic chips can be configured for a large class of behavior modes. Our results are pivotal for the development of new brain-inspired technologies" says Giacomo Indiveri, a professor of Neuroinformatics at the University of Zurich and ETH Zurich.
While the chips only contain 4000 “neurons,” a pittance when compared to the billions in our own brains, everything has to start somewhere, and the chip could prove instrumental in developing technology capable of mimicking or even surpassing human neuroprocessing.
If neuromorphic chips could be scaled up, and artificial neurons be built to contain millions of connections apiece, I wonder if chips like these could be used to create true artificial intelligence systems, or even electronic repositories for own brains? In the future it just might be possible to download your own brain into a neuromorphic computing environment, where your consciousness might continue well beyond its usual expiration date.
If you’d like to get away from my sci-fi rambling and look deeper into this subject, the Zurich team’s work was recently published in the Proceedings of the Nation Academy of Sciences of the US.
Images Courtesy of WikiMedia Commons