Research could Accelerate AI Systems to Compute Faster Than Human Brain

Two State University of New York Polytechnic Institute (SUNY Poly) researchers have been awarded $900,000 in funding from the Rome-based Air Force Research Laboratory (AFRL) to conduct a study on brain-inspired (neuromorphic) computing systems comprised of quantum devices operating at cryogenic (below −450 °F) temperatures.

Satyavolu Papa Rao, associate vice president for research and adjunct professor of nanoscience, and Nathaniel Cady, professor of nanobioscience, will work to address the stumbling blocks in all-electronic implementations of neuromorphic computing by research and development of the critical elements of superconducting optoelectronics at the 300-mm scale.

Satyavolu Papa Rao, associate vice president for research and adjunct professor of nanoscience at the State University of New York Polytechnic Institute. Courtesy of SUNY.

The brain-inspired infrastructure will use ultrafast, extremely energy-efficient Josephson junctions, which consist of two superconducting materials and a thin nonsuperconducting material in between.

The Josephson junctions will be combined with silicon-based infrared photon emitters, which generate light pulses that allow a given neuron to communicate with many downstream neurons. This arrangement mimics how the human brain works by sending and receiving ultrashort electrical pulses that it uses to simultaneously store and process information.

Research will be conducted in SUNY Poly’s 300-mm wafer fabrication facility using the same tool platforms on which advanced computer chips are built. This research, the scientists said, can accelerate the development of large-scale, fab-friendly superconducting optoelectronic systems that could compute 30,000× faster than the human brain, but at the same level of energy efficiency.