Our brain is a marvel of efficiency. It only needs as much energy as a light bulb to process information. Scientists from Jülich are researching how this works. With this knowledge, they are developing novel and particularly energy-efficient computer technologies.

The brain serves as a model for constructing computers that are fast and energy-efficient at the same time. For certain tasks, conventional computers still require a thousand times more energy than our brain – for example, for pattern recognition, i.e. recognizing a familiar face in a crowd or a misspelled word in a text.

Neuromorphic technologies could fundamentally change artificial intelligence.

At Jülich, scientists from a range of disciplines collaborate on the development of technologies for neuromorphic computing. Hardware and software engineers, semiconductor experts, theorists, but also neuroscientists are involved – Forschungszentrum Jülich brings together a wide range of scientific expertise. We are convinced that interdisciplinary research is important – and our success proves us right.

Our pioneering work in this field is also having an impact in the region: the development of neuromorphic computer chips is an economic advantage for the Rhenish mining area.

Computer chips modelled on synapses

86 billion neurons and even more synapses – the human brain is incredibly complex and its secrets are not yet fully understood. But this much is known: the brain processes and stores signals at all levels simultaneously in one place. In conventional computers, by contrast, memory and processing units are separate. Data are exchanged between the processor and the hard drive or working memory, and processed one after the other. This requires a lot of energy. The approach taken in neuromorphic computing is therefore called “in-memory computing” – computing directly in the memory.

Jülich researchers are developing chips that carry memristors (derived from the terms “memory” and “resistor”). They are structured like synapses, which transmit signals between neurons in the brain. By adjusting the electrical resistance during signal transduction, information can be stored and processed on memristors. The special feature is that these innovative chips, which can be miniaturized down to the nanoscale, can work together with conventional microelectronics. Machine learning and artificial intelligence in particular could benefit from such modular computer systems in the future.

NEUROTEC

The NEUROTEC project, in which Forschungszentrum Jülich is collaborating with RWTH Aachen University and companies in the region, aims to create a technological basis for neuro-inspired AI hardware from Europe.

JUNCA

The Jülich Neuromorphic Computing Alliance (JUNCA) brings together scientists from the field of neuromorphic computing. The exchange of knowledge and data helps to accelerate research.

EBRAINS

The digital research infrastructure EBRAINS (European Brain Infrastructures) pools the constantly growing knowledge about the brain from different disciplines in one place. It is used by scientists around the world.

The Jülich Supercomputing Centre (JSC) is where all the threads come together. It is already developing modular concepts that combine conventional computers with neuromorphic systems and quantum computers. This way, a supercomputer is being created according to the building-block principle, which will help us to achieve new levels of computing power and energy efficiency.

Neuromorphic computing is useful for a wide range of applications. Potential use cases range from autonomous driving and learning industrial control systems to the construction of intelligent and self-sufficient implants. In addition, supercomputers that function like our brain can advance brain research. Scientists use them to simulate data processing in the brain and thus unlock its secrets. The insights gained are then used to design new hardware components for neuromorphic computers. In this way, the different disciplines can benefit from each other. At Jülich, they even work side by side on the same campus.