Quantum-limited stochastic optical neural networks operating at a few quanta per activation
arXiv:2307.15712 · doi:10.1038/s41467-024-55220-y
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, 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.
65 pages, 28 figures
References in corpus (24)
- Deep Learning with Coherent Nanophotonic Circuits
- Training and Operation of an Integrated Neuromorphic Network Based on Metal-Oxide Memristors
- All-Optical Machine Learning Using Diffractive Deep Neural Networks
- Parallel convolution processing using an integrated photonic tensor core
- Quantum Circuit Learning
- 11 TeraFLOPs per second photonic convolutional accelerator for deep learning optical neural networks
- Universal Linear Optics
- Convolutional Networks for Fast, Energy-Efficient Neuromorphic Computing
- Deep physical neural networks enabled by a backpropagation algorithm for arbitrary physical systems
- Challenges and Opportunities in Quantum Machine Learning
- Single-chip photonic deep neural network for instantaneous image classification
- Large-scale neuromorphic optoelectronic computing with a reconfigurable diffractive processing unit
- Vowel recognition with four coupled spin-torque nano-oscillators
- Binary Neural Networks: A Survey
- Accurate deep neural network inference using computational phase-change memory
- Wave Physics as an Analog Recurrent Neural Network
- An optical neural network using less than 1 photon per multiplication
- Quantum Shot Noise
- Delocalized Photonic Deep Learning on the Internet's Edge
- Quantum neuromorphic computing
- Fully reconfigurable coherent optical vector-matrix multiplication
- Biasing the quantum vacuum to control macroscopic probability distributions
- Understanding and mitigating noise in trained deep neural networks
- Quantifying power use in silicon photonic neural networks