Superconducting Optoelectronic Neurons III: Synaptic Plasticity

TL;DR

This paper presents superconducting optoelectronic neuron circuits capable of dynamic synaptic weight reconfiguration, including binary, multi-stable, supervised, and unsupervised learning, with detailed circuit designs and models for various plasticity types.

Contribution

It introduces novel superconducting flux storage circuits for synaptic plasticity, enabling multiple stable weights and learning rules in optoelectronic neurons.

Findings

Circuits for hundreds of synaptic weights are designed.

Supervised learning circuits using flux pulses are demonstrated.

Models for Hebbian and anti-Hebbian unsupervised learning are presented.

Abstract

As a means of dynamically reconfiguring the synaptic weight of a superconducting optoelectronic loop neuron, a superconducting flux storage loop is inductively coupled to the synaptic current bias of the neuron. A standard flux memory cell is used to achieve a binary synapse, and loops capable of storing many flux quanta are used to enact multi-stable synapses. Circuits are designed to implement supervised learning wherein current pulses add or remove flux from the loop to strengthen or weaken the synaptic weight. Designs are presented for circuits with hundreds of intermediate synaptic weights between minimum and maximum strengths. Circuits for implementing unsupervised learning are modeled using two photons to strengthen and two photons to weaken the synaptic weight via Hebbian and anti-Hebbian learning rules, and techniques are proposed to control the learning rate. Implementation of short-term plasticity, homeostatic plasticity, and metaplasticity in loop neurons is discussed.