An all-optical convolutional neural network for image identification

TL;DR

This paper presents an all-optical convolutional neural network that achieves high accuracy and unprecedented energy efficiency for image classification, leveraging optical diffraction to bypass nonlinear activation challenges.

Contribution

The authors demonstrate a novel all-optical CNN architecture that eliminates the need for explicit optical nonlinearities, enabling ultrafast and energy-efficient image classification.

Findings

Achieved 86.8% accuracy on MNIST dataset.

Attained 94.8% accuracy on a ten-class gesture dataset.

Reported the highest energy efficiency among CNN hardware.

Abstract

In modern artificial intelligence, convolutional neural networks (CNNs) have become a cornerstone for visual and perceptual tasks. However, their implementation on conventional electronic hardware faces fundamental bottlenecks in speed and energy efficiency due to resistive and capacitive losses. Photonic alternatives offer a promising route, yet the difficulty of realizing optical nonlinearities has prevented the realization of all-optical CNNs capable of end-to-end image classification. Here, we demonstrate an all-optical CNN that bypasses the need for explicit optical nonlinear activations. Our architecture comprises a single spatial-differentiation convolutional stage--using 24 directional kernels spanning 360{\deg}, along with a mean-filtering kernel--followed by a diffractive fully-connected layer. The directional convolution enhances feature selectivity, suppresses noise and crosstalk, and simplifies the classification task, allowing the weak nonlinearity inherent in optical diffraction to achieve high accuracy. We report experimentally classification accuracies of 86.8% on handwritten digits (MNIST) and 94.8% on a ten-class gesture dataset. The system delivers a computational throughput of 1.13X10^5 tera-operations per second (TOPS) and an energy efficiency of 1.51X10^3 TOPS/W--the highest reported among CNN hardware--with the potential to improve by a further 5-6 orders of magnitude using nanosecond-scale detectors. This work establishes a scalable pathway toward ultralow-latency, ultralow-energy vision processing for real-time intelligent systems.