Trainable dynamical masking for readout-free optical computing

TL;DR

This paper introduces a trainable dynamical masking technique using optical communication devices to enhance optical computing, enabling efficient regression and time series prediction without traditional readout layers.

Contribution

It presents a novel approach to optical computing by integrating trainable dynamical masks with off-the-shelf optical devices, replacing traditional readout layers.

Findings

Effective for regression tasks

Successful time series prediction

Utilizes existing optical communication hardware

Abstract

Nonlinear systems, transforming an input signal into a high-dimensional output feature space, can be used for non-conventional computing. This approach, however, requires a change of system parameters during training rather than coefficients in a software program. We propose here to use available off-the-shelf high-speed optical communication devices and technologies to implement a trainable dynamical mask in addition to or even instead of the traditional readout layer for extreme learning machine-based computing. The computational potential of the proposed approach is demonstrated with both regression and time series prediction tasks.