NIST researchers have developed a superconducting circuit that mimics the behavior of biological synapses, the connections between neurons in the brain. The circuit uses single photons to transmit signals, similar to how neurons communicate with each other. This breakthrough could lead to artificial neural systems that operate 100,000 times faster than the human brain.
The key innovation is a superconducting single-photon detector that works in conjunction with a Josephson junction, a component that produces small voltage pulses called fluxons when a certain threshold is reached. By applying an external bias, researchers can tune the amount of current added to the loop per photon, effectively setting the synaptic weight.
The stored current serves as short-term memory, with a duration that can be varied from hundreds of nanoseconds to milliseconds. This allows the hardware to be matched to problems occurring at different time scales. The next step is to combine these synapses with on-chip light sources to demonstrate full superconducting optoelectronic neurons.
The potential impact of this work is significant, as it could enable more efficient and powerful neuromorphic computing systems. The research was funded in part by the Defense Advanced Research Projects Agency.
Keywords: photons, superconducting, neurons, synapses, Josephson
