Increasing AI computing power and high energy requirements call for new computer architectures. Researchers at Forschungszentrum Jülich are working on neuromorphic computing – a neuro-inspired AI technology that utilizes the properties of the biological brain to make AI applications more energy-efficient and faster.
Increasing AI computing power and the enormous power requirements of artificial intelligence pose major challenges for existing computer systems. At the same time, previous efficiency improvements are insufficient. The biological brain demonstrates that information-processing systems can operate significantly more efficiently.
Researchers at the Peter Grünberg Institute at Forschungszentrum Jülich are therefore developing neuro-inspired AI technologies based on neuromorphic computing. This approach utilizes properties of the biological brain for the design of hardware and software in computing systems.
The technology is based on memristor memory elements and other innovative electronic components that mimic neurons and synapses. The aim is to implement AI algorithms through a co-design of hardware and software.
Neuromorphic computing is expected to enable AI tasks with ten to a hundred times greater efficiency. This involves the industrialization of neuromorphic chip design and co-integrated component manufacturing, as well as the acceleration of machine learning and AI algorithms. Potential applications include transport, IoT sensor technology, healthcare and robotics.
Further information: https://www.fz-juelich.de/en/pgi/research/neuromorphic-computing
Wissenschaftlicher Koordinator PGI-SO
