Low-power scalable multilayer optoelectronic neural networks enabled with incoherent light
Optical approaches have made great strides towards the goal of high-speed, energy-efficient computing necessary for modern deep learning and AI applications. Read-in and read-out of data, however, limit the overall performance of existing approaches. This study introduces a multilayer optoelectronic computing framework that alternates between optical and optoelectronic layers to implement matrix-vector multiplications and rectified linear functions, respectively. Our framework is designed for real-time, parallelized operations, leveraging 2D arrays of LEDs and photodetectors connected via independent analog electronics. We experimentally demonstrate this approach using a system with a three-layer network with two hidden layers and operate it to recognize images from the MNIST database with a recognition accuracy of 92% and classify classes from a nonlinear spiral data with 86% accuracy. By implementing multiple layers of a deep neural network simultaneously, our approach significantly reduces the number of read-ins and read-outs required and paves the way for scalable optical accelerators requiring ultra low energy.
| Author(s): | Song, Alexander and Kottapalli, Sai Nikhilesh Murty and Goyal, Rahul and Schoelkopf, Bernhard and Fischer, Peer |
| Journal: | Nature Communications |
| Volume: | 15 |
| Pages: | 10692 |
| Year: | 2024 |
| BibTeX Type: | Article (article) |
| DOI: | https://doi.org/10.1038/s41467-024-55139-4 |
| State: | Published |
| URL: | https://www.nature.com/articles/s41467-024-55139-4 |
| Article Number: | 10692 |
| Digital: | True |
BibTeX
@article{Songetal24,
title = {Low-power scalable multilayer optoelectronic neural networks enabled with incoherent light},
journal = {Nature Communications},
abstract = {Optical approaches have made great strides towards the goal of high-speed, energy-efficient computing necessary for modern deep learning and AI applications. Read-in and read-out of data, however, limit the overall performance of existing approaches. This study introduces a multilayer optoelectronic computing framework that alternates between optical and optoelectronic layers to implement matrix-vector multiplications and rectified linear functions, respectively. Our framework is designed for real-time, parallelized operations, leveraging 2D arrays of LEDs and photodetectors connected via independent analog electronics. We experimentally demonstrate this approach using a system with a three-layer network with two hidden layers and operate it to recognize images from the MNIST database with a recognition accuracy of 92% and classify classes from a nonlinear spiral data with 86% accuracy. By implementing multiple layers of a deep neural network simultaneously, our approach significantly reduces the number of read-ins and read-outs required and paves the way for scalable optical accelerators requiring ultra low energy.},
volume = {15},
pages = {10692},
year = {2024},
author = {Song, Alexander and Kottapalli, Sai Nikhilesh Murty and Goyal, Rahul and Schoelkopf, Bernhard and Fischer, Peer},
doi = {https://doi.org/10.1038/s41467-024-55139-4},
url = {https://www.nature.com/articles/s41467-024-55139-4}
}