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

Abstract Energy efficiency in computation is ultimately limited by noise, with quantum limits setting the fundamental noise floor. Analog physical neural networks hold promise for improved energy efficiency compared to digital electronic neural networks. However, they are typically operated in a rel...

Full description

Saved in:
Bibliographic Details
Main Authors: Shi-Yuan Ma, Tianyu Wang, Jérémie Laydevant, Logan G. Wright, Peter L. McMahon
Format: Article
Language:English
Published: Nature Portfolio 2025-01-01
Series:Nature Communications
Online Access:https://doi.org/10.1038/s41467-024-55220-y
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850048901314248704
author Shi-Yuan Ma
Tianyu Wang
Jérémie Laydevant
Logan G. Wright
Peter L. McMahon
author_facet Shi-Yuan Ma
Tianyu Wang
Jérémie Laydevant
Logan G. Wright
Peter L. McMahon
author_sort Shi-Yuan Ma
collection DOAJ
description Abstract Energy efficiency in computation is ultimately limited by noise, with quantum limits setting the fundamental noise floor. Analog physical neural networks hold promise for improved energy efficiency compared to digital electronic neural networks. However, they are typically operated in a relatively high-power regime so that the signal-to-noise ratio (SNR) is large (>10), and the noise can be treated as a perturbation. We study optical neural networks where all layers except the last are operated in the limit that each neuron can be activated by just a single photon, and as a result the noise on neuron activations is no longer merely perturbative. We show that by using a physics-based probabilistic model of the neuron activations in training, it is possible to perform accurate machine-learning inference in spite of the extremely high shot noise (SNR  ~ 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 just 0.038 photons per multiply-accumulate (MAC) operation. Our physics-aware stochastic training approach might also prove useful with non-optical ultra-low-power hardware.
format Article
id doaj-art-a11fbb927dd348019f00c95e188739fd
institution DOAJ
issn 2041-1723
language English
publishDate 2025-01-01
publisher Nature Portfolio
record_format Article
series Nature Communications
spelling doaj-art-a11fbb927dd348019f00c95e188739fd2025-08-20T02:53:50ZengNature PortfolioNature Communications2041-17232025-01-0116111210.1038/s41467-024-55220-yQuantum-limited stochastic optical neural networks operating at a few quanta per activationShi-Yuan Ma0Tianyu Wang1Jérémie Laydevant2Logan G. Wright3Peter L. McMahon4School of Applied and Engineering Physics, Cornell UniversitySchool of Applied and Engineering Physics, Cornell UniversitySchool of Applied and Engineering Physics, Cornell UniversitySchool of Applied and Engineering Physics, Cornell UniversitySchool of Applied and Engineering Physics, Cornell UniversityAbstract Energy efficiency in computation is ultimately limited by noise, with quantum limits setting the fundamental noise floor. Analog physical neural networks hold promise for improved energy efficiency compared to digital electronic neural networks. However, they are typically operated in a relatively high-power regime so that the signal-to-noise ratio (SNR) is large (>10), and the noise can be treated as a perturbation. We study optical neural networks where all layers except the last are operated in the limit that each neuron can be activated by just a single photon, and as a result the noise on neuron activations is no longer merely perturbative. We show that by using a physics-based probabilistic model of the neuron activations in training, it is possible to perform accurate machine-learning inference in spite of the extremely high shot noise (SNR  ~ 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 just 0.038 photons per multiply-accumulate (MAC) operation. Our physics-aware stochastic training approach might also prove useful with non-optical ultra-low-power hardware.https://doi.org/10.1038/s41467-024-55220-y
spellingShingle Shi-Yuan Ma
Tianyu Wang
Jérémie Laydevant
Logan G. Wright
Peter L. McMahon
Quantum-limited stochastic optical neural networks operating at a few quanta per activation
Nature Communications
title Quantum-limited stochastic optical neural networks operating at a few quanta per activation
title_full Quantum-limited stochastic optical neural networks operating at a few quanta per activation
title_fullStr Quantum-limited stochastic optical neural networks operating at a few quanta per activation
title_full_unstemmed Quantum-limited stochastic optical neural networks operating at a few quanta per activation
title_short Quantum-limited stochastic optical neural networks operating at a few quanta per activation
title_sort quantum limited stochastic optical neural networks operating at a few quanta per activation
url https://doi.org/10.1038/s41467-024-55220-y
work_keys_str_mv AT shiyuanma quantumlimitedstochasticopticalneuralnetworksoperatingatafewquantaperactivation
AT tianyuwang quantumlimitedstochasticopticalneuralnetworksoperatingatafewquantaperactivation
AT jeremielaydevant quantumlimitedstochasticopticalneuralnetworksoperatingatafewquantaperactivation
AT logangwright quantumlimitedstochasticopticalneuralnetworksoperatingatafewquantaperactivation
AT peterlmcmahon quantumlimitedstochasticopticalneuralnetworksoperatingatafewquantaperactivation