Superconducting Optoelectronic Circuits for Neuromorphic Computing

Abstract

Neural networks have proven effective for solving many difficult computational problems. Implementing complex neural networks in software is very computationally expensive. To explore the limits of information processing, it will be necessary to implement new hardware platforms with large numbers of neurons, each with a large number of connections to other neurons. Here we propose a hybrid semiconductor-superconductor hardware platform for the implementation of neural networks and large-scale neuromorphic computing. The platform combines semiconducting few-photon light-emitting diodes with superconducting-nanowire single-photon detectors to behave as spiking neurons. These processing units are connected via a network of optical waveguides, and variable weights of connection can be implemented using several approaches. The use of light as a signaling mechanism overcomes fanout and parasitic constraints on electrical signals while simultaneously introducing physical degrees of freedom which can be employed for computation. The use of supercurrents achieves the low power density necessary to scale to systems with enormous entropy. The proposed processing units can operate at speeds of at least 20 MHz with fully asynchronous activity, light-speed-limited latency, and power densities on the order of 1 mW/cm 2 for neurons with 700 connections operating at full speed at 2 K. The processing units achieve an energy efficiency of ≈ 20 aJ per synapse event. By leveraging multilayer photonics with deposited waveguides and superconductors with feature sizes > 100 nm, this approach could scale to systems with massive interconnectivity and complexity for advanced computing as well as explorations of information processing capacity in systems with an enormous number of information-bearing microstates.

Active material, optical mode and cavity impact on nanoscale electro-optic modulation performance

“…Thus, if the entire system would be cooled to cryogenic levels already, the energy per bit could be lowered by about a factor of ~10 × [61]. Indeed cooling devices to cryogenic temperatures could actually become an interesting option in single and few photon systems [62].…”

Fluxonic Processing of Photonic Synapse Events

“…For hardware to be efficient for neural information processing, synaptic and dendritic operations must be efficiently manifest in constituent devices. We have argued elsewhere that light is promising for communication in neural systems because it enables the fan-out and energy efficiency necessary for large neural systems, and that utilization of superconducting single-photon detectors enables communication at the lowest possible light levels [11]. Subsequent work considered specific synaptic and neuronal circuits suitable for point-neuron behavior, introducing circuits capable of transducing single-photon communication events to the electronic domain for subsequent information processing [12,13].…”

Microresonator-enhanced, Waveguide-coupled Emission from Silicon Defect Centers for Superconducting Optoelectronic Networks

“…Extending the application areas of hardware neuromorphics will require new physical platforms, such as those based on photonics, analog electronics, or Josephson junctions. Superconducting optoelectronic neural networks (SOENs) offer a route to hardware neuromorphic systems with the largest scales considered to date [1]. In the SOEN architecture, light is used for communication in order to avoid bandwidth/distance/dissipation tradeoffs affecting both analog and digital electronic interconnects.…”