This is a Research and Innovation ActionÂ of the EU Horizon 2020 Programme.
We propose to fabricate a chip implementing a neuromorphic architecture that supports state-of-the-art machine learningÂ and spike-based learning mechanisms. With respect to its physical architecture this chip will feature an ultra-lowÂ power, scalable, and highly configurable neural architecture that will deliver a gain of a factor 50x in power consumption onÂ selected applications compared to conventional digital solutions; and a monolithically integrated 3D technology in Fully-Depleted Silicon on Insulator (FDSOI) at 28nm design rules with integrated Resistive Random Access Memory (RRAM)Â synaptic elements.
We will complete this vision and develop complementary technologies that will allow to address the full spectrum ofÂ applications from mobile/autonomous objects to high performance computing coprocessing, by realising (1) a technology toÂ implement on-chip learning, using native adaptive characteristics of electronic synaptic elements; and (2) a scalable platformÂ to interconnect multiple neuromorphic processor chips to build large neural processing systems.
The neuromorphic computing system will be developed jointly with advanced neural algorithmsÂ and computational architectures for online adaptation, learning, and high-throughput on-line signalÂ processing, delivering:
- an ultra-low power massively parallel non-Von Neumann computing platform with non-volatile nano-scale devices thatÂ support on-line learning mechanisms;
- a programming toolbox of algorithms and data structures tailored to the specific constraints and opportunities of theÂ physical architecture;Â
- an array of fundamental application demonstrations instantiating the basic classes of signal processing tasks.Â
The neural chip will validate the concept and be a first step to develop a European technology platform addressing fromÂ ultra-low power data processing in autonomous systems (Internet of Things) to energy-efficient large data processing inÂ servers and networks.