Compressed Meta-Optical Encoder for Image Classification

TL;DR

This paper presents a hybrid optical-electronic neural network that uses a meta-optic to perform convolution, significantly reducing computational complexity and power consumption while maintaining high accuracy in image classification.

Contribution

The work introduces a novel hybrid optical-electronic CNN architecture with a meta-optic for convolution, achieving over two orders of magnitude reduction in operations and power use.

Findings

Achieved over 93% accuracy on MNIST dataset.

Reduced multiply-accumulate operations from 17M to 86K.

Demonstrated over two orders of magnitude decrease in latency and power consumption.

Abstract

Optical and hybrid convolutional neural networks (CNNs) recently have become of increasing interest to achieve low-latency, low-power image classification and computer vision tasks. However, implementing optical nonlinearity is challenging, and omitting the nonlinear layers in a standard CNN comes at a significant reduction in accuracy. In this work, we use knowledge distillation to compress modified AlexNet to a single linear convolutional layer and an electronic backend (two fully connected layers). We obtain comparable performance to a purely electronic CNN with five convolutional layers and three fully connected layers. We implement the convolution optically via engineering the point spread function of an inverse-designed meta-optic. Using this hybrid approach, we estimate a reduction in multiply-accumulate operations from 17M in a conventional electronic modified AlexNet to only 86K in the hybrid compressed network enabled by the optical frontend. This constitutes over two orders of magnitude reduction in latency and power consumption. Furthermore, we experimentally demonstrate that the classification accuracy of the system exceeds 93% on the MNIST dataset.