Nano-Opto-Electro-Mechanical Integrated Oscillator Arrays for Energy-Efficient Physical Reservoir Computing (NOEMIA)
Lead Research Organisation:
University of Southampton
Department Name: Sch of Electronics and Computer Sci
Abstract
A significant percentage of the global energy production is consumed by Information and Communications Technology (ICT), and the consumption will be further accelerated with rapid and broad adoption of Artificial Intelligence (AI) where demand on complex machine learning models with high power consumption is increasing.
The NOEMIA project aims to develop a fundamentally novel energy-efficient hardware alternative for machine learning by using Nano-Opto-Electro-Mechanical Systems (NOEMS) technology. Within the project timeline, NOEMIA targets to deliver a world-first Nano-Opto-Electro-Mechanical (NOEM) coupled oscillator array integrated with GeSn photodetectors fabricated via silicon-compatible technologies. Active integration of silicon photonics components with nanomechanical oscillators will allow us to detect sub-nW oscillation with excellent sensitivity and with minimal energy loss. The dynamic behaviour and scalability of the systems will be explored theoretically and device and material-level system tunability will be investigated towards the application for physical reservoir computing, a fast, simple, and energy-efficient option of machine learning.
The NOEMIA project (1.16M EUR in total) is approved in the EU CHIST-ERA scheme, and the project consortium consists of 5 institutions from 5 different countries with complementary expertise. The project lead, University of Southampton (UOS) in the UK offers strong expertise of nanodevice fabrication and advanced nanophotonics, and Institut d'électronique de microélectronique et de nanotechnologie (IEMN) in France provides skills of MEMS technology and opto-electro-mechanical characterisation. National Taiwan University (NTU) in Taiwan contributes with their Si photonics and GeSn photodetector expertise, and laser annealing and material modification are conducted by Riga Technical University (RTU) in Latvia. The Institute of Thermomechanics of the Czech Academy of Sciences (CAS) in Czech Republic leads theoretical investigation of nonlinear oscillators.
Successful demonstration of novel NOEM integrated oscillator arrays by the unique and interdisciplinary NOEMIA consortium will make a significant impact on the development of next generation AI hardware, one of the most exciting research fields in ICT.
The NOEMIA project aims to develop a fundamentally novel energy-efficient hardware alternative for machine learning by using Nano-Opto-Electro-Mechanical Systems (NOEMS) technology. Within the project timeline, NOEMIA targets to deliver a world-first Nano-Opto-Electro-Mechanical (NOEM) coupled oscillator array integrated with GeSn photodetectors fabricated via silicon-compatible technologies. Active integration of silicon photonics components with nanomechanical oscillators will allow us to detect sub-nW oscillation with excellent sensitivity and with minimal energy loss. The dynamic behaviour and scalability of the systems will be explored theoretically and device and material-level system tunability will be investigated towards the application for physical reservoir computing, a fast, simple, and energy-efficient option of machine learning.
The NOEMIA project (1.16M EUR in total) is approved in the EU CHIST-ERA scheme, and the project consortium consists of 5 institutions from 5 different countries with complementary expertise. The project lead, University of Southampton (UOS) in the UK offers strong expertise of nanodevice fabrication and advanced nanophotonics, and Institut d'électronique de microélectronique et de nanotechnologie (IEMN) in France provides skills of MEMS technology and opto-electro-mechanical characterisation. National Taiwan University (NTU) in Taiwan contributes with their Si photonics and GeSn photodetector expertise, and laser annealing and material modification are conducted by Riga Technical University (RTU) in Latvia. The Institute of Thermomechanics of the Czech Academy of Sciences (CAS) in Czech Republic leads theoretical investigation of nonlinear oscillators.
Successful demonstration of novel NOEM integrated oscillator arrays by the unique and interdisciplinary NOEMIA consortium will make a significant impact on the development of next generation AI hardware, one of the most exciting research fields in ICT.
