University of Tennessee, Knoxville researchers have developed a neuromorphic chip designed for intelligent computers that are structured to discover patterns through probabilities and association in order to help with decision making.
The researchers used off-the-shelf, reprogrammable circuits called field programmable gate arrays (FPGAs) to simulate the way neurons and synapses in a brain operate. FPGAs excel at performing tasks and can be easily reprogrammed for other applications. "We believe our architectures are particularly amenable for supercomputing applications because of their programmability," the researchers say. They also are studying how to swap out FPGAs with memristors, which can retain data and are considered a replacement for dynamic random-access memory.
There are disadvantages to using FPGAs for a practical brain model, however. Reprogramming FPGAs requires bringing them offline, which can disrupt the execution of tasks. In addition, FPGA's cannot be primary chips that boot systems, as they are mostly being used as co-processors that can be power hungry. The researchers believe the chip architecture is more important than the type of chip, and more chip prototypes based on the architecture will be made available to other researchers.
From IDG News Service
View Full Article
Abstracts Copyright © 2016 Information Inc., Bethesda, Maryland, USA