欧盟-韩国合作项目为边缘计算提供节能的神经形态二维设备.pdf

1、ENERGIZEEnergy-efficient Neuromorphic 2d Devices And Circuits For Edge AI Computing Funded by the European UnionENERGIZE EU-ROK collaborative project to enable energy efficient neuromorphic two-dimensional devices for edge computingDmitry Chigrin,AMO GmbH,Aachen,GermanyENERGIZEEnergy-efficient Neuro

2、morphic 2d Devices And Circuits For Edge AI Computing Funded by the European UnionWhat2ENERGIZEs vision is to leverage the potential of wafer-scale,2D materials-based neural networks to develop energy-efficient neuromorphic devices and circuits for edge AI computing.ENERGIZEEnergy-efficient Neuromor

3、phic 2d Devices And Circuits For Edge AI Computing Funded by the European UnionWhy3Source:SRC Decadal Plan,2020 Machine Learning and Artificial Intelligence need new hardware Neuromorphic ComputingVon-Neumann bottleneckProcessorMemorySebastian et al.,Nat.Nanotechnol.,116,2020ENERGIZEEnergy-efficient

4、 Neuromorphic 2d Devices And Circuits For Edge AI Computing Funded by the European UnionWhoAMO GmbH(AMO)Universita di Pisa(UNIPI)Universidad de Granada(UGR)Ecole Polytechnique Federale de Lausanne(EPFL)Sungkyunkwan University(SKKU)Korea University(KU)Gwangju Institute of Science and Technology(GIST)

5、Sogang University(SGU)4ENERGIZEEnergy-efficient Neuromorphic 2d Devices And Circuits For Edge AI Computing Funded by the European UnionSimulation and Modelling5Objective:To provide a multiscale simulation approach for the study of devices for neuromorphic electronics,spanning from atomistic to large

6、 scale circuit emulations.Partners:UNIPI:Atomistic simulations of 2DMs,multiscale transport simulation of devices up to circuit level.UGR:Simulation of 2DMs,devices and circuits,compact modelling of devices and validation.ENERGIZEEnergy-efficient Neuromorphic 2d Devices And Circuits For Edge AI Comp