Perceptrons with Hebbian learning based on wave ensembles in plastic potentials

TL;DR

This paper proposes a hardware perceptron model utilizing wave superpositions in patterned potentials, enabling optical neural network implementation across various optical systems through Hebbian learning.

Contribution

It introduces a novel scheme for perceptrons using wave ensembles and Hebbian learning, adaptable to multiple optical hardware platforms.

Findings

Potential shapes derived from Hebbian learning enable optical perceptron implementation.

Superposition of Schrödinger waves allows interconnection in optical neural networks.

Applicable to diverse optical systems including lithography-patterned and exciton-polariton systems.

Abstract

A general scheme to realize a perceptron for hardware neural networks is presented, where multiple interconnections are achieved by a superposition of Schrodinger waves. Spatially patterned potentials process information by coupling different points of reciprocal space. The necessary potential shape is obtained from the Hebbian learning rule, either through exact calculation or construction from a superposition of known optical inputs. This allows implementation in a wide range of compact optical systems, including: 1) any non-linear optical system; 2) optical systems patterned by optical lithography; and 3) exciton-polariton systems with phonon or nuclear spin interactions.