Developing hardware and algorithms for computers modeled on the brain - that is the goal of the NEUROTEC project, which is now entering its second funding phase.
As part of the project, researchers are working on practical demonstrators to show how much more efficient neuro-inspired AI will be. The project network, which is coordinated by Forschungszentrum Jülich, has received total funding of around 36 million euros for the next five years from the German Federal Ministry of Education and Research (BMBF). The research center will receive around 21 million euros of this.
The second funding phase is part of the Coal Regions Structural Strengthening Act passed by the German Bundestag and Bundesrat and the PLUS emergency program of the state of North Rhine-Westphalia to support structural change following the phase-out of coal.
»By continuing the project, we want to make an important contribution to the development of the region around Jülich and Aachen into a globally visible center for neuro-inspired information technologies. We are pleased to be able to use an interdisciplinary approach to lay the foundation for leapfrog innovations in order to create opportunities for companies with high added value and sustainable jobs in the Rhineland region,« explains Prof. Dr.-Ing. Wolfgang Marquardt, Chairman of the Board of Directors of Forschungszentrum Jülich.
The interdisciplinary NEUROTEC project brings together the expertise of Forschungszentrum Jülich and RWTH Aachen University, which are working with high-tech companies to develop new materials and electronic components for neuroinspired hardware. Numerous regionally based companies such as AIXTRON, AMO, aixACCT Systems, SURFACE systems+technology and AMOtronics are involved in the project.
The way the brain works serves as a model for innovative computer architectures, so-called neuromorphic computers. This is because the human brain works extremely efficiently. For certain tasks, such as recognizing patterns, it requires thousands of times less energy than conventional digital computers. AI and machine learning in particular could benefit from neuromorphic systems. This is because these are constructed similarly to the biological networks they emulate. Other future applications include wherever data needs to be processed in real time, such as for autonomous driving or Industry 4.0.
Neuromorphic computing based on memristors
NEUROTEC focuses on the smallest electronic components that can be used for neuromorphic computing, so-called memristive components, which can be connected to conventional microelectronics. The project uses coating systems, measurement technology and software from high-tech suppliers in the region to manufacture them. Memristors have similar properties to the synapses in the human brain. They can store and simultaneously process information via an adjustable resistance value. As far as the fundamental material understanding of these switching elements is concerned, the alliance of Forschungszentrum Jülich and RWTH Aachen University is considered a world leader.
»The second phase of the project will primarily focus on developing demonstrators to prove the technology and function of the neuromorphic hardware and software architectures,« explains Prof. Rainer Waser of Forschungszentrum Jülich and RWTH Aachen University, who is coordinating the NEUROTEC project. »To do this, we can draw on the expanded system infrastructure and concepts that have been developed in the first phase of the project since 2019,« says Waser, who received the Leibniz Prize in 2014 for his research.
The collaborative project will receive personnel reinforcement in its second phase from Prof. Dr. John Paul Strachan and Prof. Dr. Emre Ozgur Neftci. The two scientists moved to Forschungszentrum Jülich in July from Hewlett-Packard Labs and the University of California and specialize in the development of hardware and software for neuromorphic systems.
Project synergies with future cluster for neuromorphic hardware
Thematic points of contact are offered by the NeuroSys future cluster, which is working on the development of neuromorphic hardware for autonomous artificial intelligence systems under the leadership of RWTH Aachen University. The goal is to develop the region into an innovation ecosystem consisting of research institutions, industrial settlements and start-ups as a world-leading location for research, development and use of European AI hardware. The results achieved in NEUROTEC will flow into NeuroSys, where they will be further driven toward application. The cluster, which is also funded by the BMBF, bundles regional competencies and infrastructures of partners from the fields of science, industry and civil society for this purpose.
»With the NEUROTEC project, we are developing technologies for a completely new generation of chips and hardware,« says Federal Research Minister Anja Karlizcec. Modeled on the human brain, the aim is to significantly increase energy efficiency for artificial intelligence applications. »I am pleased that the second phase of the project is now starting and will be funded with around 36 million euros by the Federal Ministry of Research.«