Time-Delayed Reservoir Computing Based on a Two-Element Phased Laser Array for Image Identification

Yu Huang,Nianqiang Li,Pei Zhou,Taiyi Chen,Yigong Yang

doi:10.1109/jphot.2021.3115598

Abstract

We report on a simple approach of time-delayed reservoir computing (RC) based on a two-element phased laser array for image identification. Here the phased laser array with optical feedback and injection is trained according to the representative characteristics extracted through histograms of oriented gradients. These characteristic vectors are multiplied by a random mask signal to form input data, which are subsequently trained in the reservoir. By optimizing the parameters of the RC, we achieve an identification accuracy of 97.44% on the MNIST dataset and 85.46% on the Fashion-MNIST dataset. These results indicate that our proposed RC indeed allows accurate classification of handwritten digit and fashion production. Moreover, we further forecast an RC scheme based on a larger-scale phased laser array, which is expected to tackle more complex tasks at a high speed. Our work offers a possibility for advanced image processing using highly integrated neuromorphic photonic systems.

Highlights

T HE CLASSIFICATION technology of static images has attracted much attention due to their broad fields of applications, such as objection detection and facial recognition in security and authentication, as well as automatic analysis of medical images and intelligent character recognition for handwritten interpretation in health sciences [1]–[5]
We utilize the two-element phased laser array as the response laser (RL) in the reservoir layer and two different tasks from different input streams can be effectively executed in parallel at the same time
We have numerically demonstrated a novel time-delayed reservoir computing (RC) scheme based on a phased laser array with optical feedback and injection

Summary

INTRODUCTION

T HE CLASSIFICATION technology of static images has attracted much attention due to their broad fields of applications, such as objection detection and facial recognition in security and authentication, as well as automatic analysis of medical images and intelligent character recognition for handwritten interpretation in health sciences [1]–[5]. Brunner and Fischer proposed and demonstrated a photonic reservoir with a diffractive optical element by using an 8×8 laser array and a spatial light modulator [29] They observed experimentally a highly nonlinear response of the phased array and realized the integrated and weighted network state needed for the RC. We propose a novel approach to processing image identification based on the time-delayed RC using two laterally coupled SLs that are coupled via evanescent fields This is the simplest laser array and it has been precisely described by leveraging a set of ordinary differential equations [25], [30]– [34]. We discuss a potential scheme that is expected to realize a larger-scale high-speed RC based on a phased laser array for intensive tasks with higher accuracy, which will be reported elsewhere

SCHEME AND THEORETICAL MODEL

RESULTS AND DISCUSSION

CONCLUSION