Quantum-noise-limited optical neural networks operating at a few quanta per activation

2023-07-28

Research Square

AI Summary

This work addresses robust inference for ultra-low-power optical neural networks operating under quantum-noise-dominated conditions. Methodologically, we propose a single-photon-level photonic neural network architecture built from linear optical components and single-photon detectors, enabling analog photonic computation. To our knowledge, this is the first demonstration of high-accuracy inference in the single-photon multiply–accumulate (MAC) regime (~1 photon/MAC). We further introduce a quantum noise model and a noise-aware backpropagation training strategy, departing from conventional high-SNR operational paradigms. Experimentally, the network achieves 98% test accuracy on MNIST at an extreme signal-to-noise ratio (SNR) of ≈1. It consumes only 0.008 photons/MAC (0.003 aJ/MAC), improving energy efficiency by over 40× compared to prior state-of-the-art. This work establishes a new paradigm for quantum-noise-tolerant, near-sensor intelligent photonic computing.

Abstract

A practical limit to energy efficiency in computation is ultimately from noise, with quantum noise [1] as the fundamental floor. Analog physical neural networks [2], which hold promise for improved energy efficiency and speed compared to digital electronic neural networks, are nevertheless typically operated in a relatively high-power regime so that the signal-to-noise ratio (SNR) is large (>10). We study optical neural networks [3] operated in the limit where all layers except the last use only a single photon to cause a neuron activation. In this regime, activations are dominated by quantum noise from the fundamentally probabilistic nature of single-photon detection. We show that it is possible to perform accurate machine-learning inference in spite of the extremely high noise (signal-to-noise ratio ~ 1). We experimentally demonstrated MNIST handwritten-digit classification with a test accuracy of 98% using an optical neural network with a hidden layer operating in the single-photon regime; the optical energy used to perform the classification corresponds to 0.008 photons per multiply–accumulate (MAC) operation, which is equivalent to 0.003 attojoules of optical energy per MAC. Our experiment also used >40× fewer photons per inference than previous state-of-the-art low-optical-energy demonstrations [4, 5] to achieve the same accuracy of >90%. Our training approach, which directly models the system’s stochastic behavior, might also prove useful with non-optical ultra-low-power hardware.

Problem

• Low energy consumption
• Optical neural networks
• Machine learning in noisy environments

Innovation

• Optical Neural Network
• Single-photon Activation
• Ultra-low Energy Consumption