Researchers are underway with revolutionary supercomputer to simulate the synapses of a human brain at full scale.
The neuromorphic supercomputer will be skilled with 228 trillion synaptic processes per second, which is on par with the assessed number of processes in the human brain.
A team of scientists at the International Centre for Neuromorphic Systems (ICNS) at Western Sydney University have called it DeepSouth IBM’s
The unbelievable computational power of the human brain can be realized in the way it achieves billion-billion mathematical operations per second by means of only 20 watts of power. DeepSouth accomplishes similar levels of parallel processing by retaining neuromorphic engineering, a design method that imitate the brain’s functioning.
Highlighting the characteristic features of DeepSouth, Professor André van Schaik, the Director of the ICNS, underlined that the supercomputer is designed with a unique purpose – to function in a manner parallel to networks of neurons, the elementary units of the human brain.
Neuromorphic systems employ interrelated artificial neurons and synapses to achieve tasks. These systems attempt to match the brain’s capability to absorb, familiarize, and process data in a highly parallel and dispersed manner.
Often practical in the field of AI and machine education, a neuromorphic structure is used with the goal of generating more effective and brain-like computing systems.
Traditional computing constructions are classically based on the von Neumann architecture, where computers are composed of distinct CPUs and memory units, where information and commands are stored in the latter.
DeepSouth can grip large quantities of data at a speedy pace while consuming significantly less power and being substantially smaller than conventional supercomputers.
Development in our understanding of how brains calculate using neurons is hampered by our incapability to simulate brain-like networks at scale. Simulating spiking neural networks on typical computers by means of Graphics Processing Units (GPUs) and multicore Central Processing Units (CPUs) is just too slow and power intensive. The new scheme will change that, Professor van Schaik said.
The team termed the supercomputer DeepSouth based on IBM’s TrueNorth system, which initiated the idea of creating computers that perform like large system of neurons, and Deep Blue, the primary computer to beat a world chess champion.
The name similarly gives a nod to where the supercomputer is situated geographically: Australia, which is located in the southern hemisphere.
The team believes DeepSouth will assist in the advancements in various fields like sensing, biomedical, robotics, space, and large-scale AI applications.
The team believes DeepSouth will also transform smart devices. This includes devices like mobile phones and sensors used in industrial and agriculture.
