Optical Convolutional Neural Networks -- Combining Silicon Photonics and Fourier Optics for Computer Vision

TL;DR

This paper introduces an optical hardware accelerator that combines silicon photonics and Fourier optics to enhance the efficiency of convolutional neural networks for computer vision tasks.

Contribution

It presents a novel proof-of-concept optical accelerator leveraging Fourier optics and silicon photonics to improve CNN performance.

Findings

Demonstrates feasibility of optical Fourier transform in CNN acceleration

Shows potential for large efficiency improvements over traditional methods

Provides a proof of concept for optical AI hardware

Abstract

The Convolutional Neural Network (CNN) is a state-of-the-art architecture for a wide range of deep learning problems, the quintessential example of which is computer vision. CNNs principally employ the convolution operation, which can be accelerated using the Fourier transform. In this paper, we present an optical hardware accelerator that combines silicon photonics and free-space optics, leveraging the use of the optical Fourier transform within several CNN architectures. The hardware presented is a proof of concept, demonstrating that this technology can be applied to artificial intelligence problems with a large efficiency boost with respect to canonical methods.