Researchers at IBM developed an innovative system that is equipped to enhance efficiency of artificial intelligence.
The new computer system developed by IBM researchers is based on concepts from the human brain that improves the performance at a much higher level as compared to conventional computers. The findings were published in the Journal of Applied Physics on August 09, 2018.
Modern computers’ architecture is based on the von Neumann architecture, which was developed in the 1940s. Using a central processer, von Neumann architecture functions and manages logic and arithmetic along with the system. The team proposed that brain-inspired computers could co-exist with processing and memory units, which is not found with conventional systems.
Abu Sebastian, lead author of the study, said, “if you look at human beings, we compute with 20 to 30 watts of power, whereas AI today is based on supercomputers which run on kilowatts or megawatts of power, in new architecture, going beyond von Neumann, memory has to play a more active role in computing.”
The team designed the system’s architecture on three levels similar to that of human brain. The first level solves state dynamics within the memory device, while the second level is akin to brain’s synaptic network structure where arrays of phase change memory (PCM) devices take place. The last level was inspired by the dynamic and stochastic nature of neurons and synapses to develop an enhanced computational substrate for spiking neural networks.
Using electrical pulses, they modulated the ratio of material in the crystalline and the amorphous phases that allowed support for electrical conductance. The researchers believe that the new architecture will help in easing the expanding data accumulated by artificial intelligence.