Artificial Intelligence Accelerators Based on Graphene Optoelectronic Devices
Authors:
Open Access
- 20 February 2021
- journal article
- research article
- Published by Wiley in Advanced Photonics Research
- Vol. 2 (6) , 2100048
Abstract
Optical and optoelectronic approaches of performing matrix–vector multiplication (MVM) operations have shown the great promise of accelerating machine learning (ML) algorithms with unprecedented performance. The incorporation of nanomaterials into the system can further improve the device and system performance thanks to their extraordinary properties, but the nonuniformity and variation of nanostructures in the macroscopic scale pose severe limitations for large‐scale hardware deployment. Here, a new optoelectronic architecture is presented, consisting of spatial light modulators and tunable responsivity photodetector arrays made from graphene to perform MVM. The ultrahigh carrier mobility of graphene, high‐power‐efficiency electro‐optic control, and extreme parallelism suggest ultrahigh data throughput and ultralow energy consumption. Moreover, a methodology of performing accurate calculations with imperfect components is developed, laying the foundation for scalable systems. Finally, a few representative ML algorithms are demonstrated, including singular value decomposition, support vector machine, and deep neural networks, to show the versatility and generality of the platform.Keywords
All Related Versions
Funding Information
- University of Utah
- National Science Foundation (2019336, 2008144)
This publication has 51 references indexed in Scilit:
- Excitation of Plasmonic Waves in Graphene by Guided-Mode ResonancesACS Nano, 2012
- Operation of Graphene Transistors at Gigahertz FrequenciesNano Letters, 2008
- Microscopic theory of the extraordinary optical transmissionNature, 2008
- Transparent, Conductive Carbon Nanotube FilmsScience, 2004
- Smart-pixel spatial light modulator for incorporation in an optoelectronic neural networkOptics Letters, 1995
- Experimental realization of any discrete unitary operatorPhysical Review Letters, 1994
- Hybrid optoelectronic neural networks using a mutually pumped phase-conjugate mirrorOptics Letters, 1991
- Two-dimensional programmable optical neural networkApplied Optics, 1989
- Optical information processing based on an associative-memory model of neural nets with thresholding and feedbackOptics Letters, 1985
- Fully parallel, high-speed incoherent optical method for performing discrete Fourier transformsOptics Letters, 1978