Researchers at NIST have developed new circuits using Josephson junctions (JJs) that enable key functions for single-photon optoelectronic neurons in neuromorphic computing. The circuits convert optical signals to electrical supercurrents, dynamically reconfigure synaptic weights based on photon detection, and initiate neuronal firing events when a current threshold is reached.
The circuits use superconducting-nanowire single-photon detectors (SNSPDs) in parallel with JJs to convert photons to supercurrents. A JJ in the integrating loop serves as a threshold detector, initiating neuronal firing when the integrated current reaches a critical level. The synaptic weight can be dynamically reconfigured by adding flux to a superconducting loop coupled to the receiver JJ.
This approach integrates photonic devices for massive connectivity with superconducting electronics for efficient memory and computation. The circuits enable single-photon detection, synaptic weight modification, and neuronal firing, solving key challenges in neuromorphic computing. Current limitations include JJ fabrication reliability and dynamic range, which are being investigated.
The technology could lead to new types of single-photon cameras and neuromorphic systems with photonic synaptic connectivity and superconducting memory. Implementation timeframes and practical applications are still under development.
Keywords: quantum, neurons, synapse, photons, circuits
