CogniGron is led by a group of world-class experts in the fields of Artificial Intelligence, Computer Science, Nanotechnology, Physics and Mathematics.

The research program hosts a globally diverse community of over 20 full professors and more than 100 researchers and students. Even if their ambitions are manifold, the aspirational vision of building an extremely power efficient supercomputer lies at the heart of what CogniGron sets out to achieve.

Rethinking technology

The rising data demands of the digital society we live in are becoming unsustainable. While our use of the cloud for streaming, smart devices and sharing platforms has surged, the computer chips that process these huge volumes of data have mostly remained the same.

New generations of laptops and smartphones are constantly being released, but their improvements in data processing capacity are modest at best. Supercomputers are evolving at a similar pace: ongoing innovation has yet to exceed refinements and gains that are only incremental.

A sustainable way forward requires new standard-setting inventions that can eclipse the highest levels of power efficiency currently available. By fundamentally rethinking the building blocks of the technology we know, CogniGron is determined to make that breakthrough a reality.

The human brain as the way forward

CogniGron specializes in neuromorphic engineering, a rapidly evolving field that develops materials resembling the human brain. Powerful supercomputers may seem impressive, but our brain is vastly superior in all aspects of data processing and memory storage, and much more energy efficient.

CogniGron’s researchers use the biological principles of the human brain to create cognitive materials: tiny parts with a mind of their own. A chip based on this approach would represent a leap forward that could skip hundreds, if not thousands, of generations of future computers based on existing technology.

Scaled Impact

The potential applications are boundless: CogniGron’s insights can help develop new technology for deep learning, self-driving cars, robotics or systems with the ability to predict natural disasters.

The experimental research done by CogniGron depends on state-of-the-art equipment and a substantial number of talented minds with the drive to revolutionize science. An anonymous donor provided the financial foundation for this project. For information on what you can do, please contact the UEF.

A computer that works like our brain, interview with Georgi Gaydadjiev

Data traffic is increasing at a rate of knots in the current digital era. In fact, it’s growing so fast that our computers can hardly keep up with the developments. In addition, huge volumes of energy are needed to process all this data. At present, IT accounts for over 10 percent of our energy consumption worldwide. If we are to future-proof society, we need an entirely new type of computer, based on completely different principles.

This is the claim being made by CogniGron, a top-level Groningen collective, funded by the Ubbo Emmius Fund. CogniGron employs 22 professors, and over 100 lecturers, PhD students, and Master’s students from all around the world. They are working together to design the energy-efficient, ultra-powerful computers of the future.

One of these professors is Georgi Gaydadjiev. He is originally from Bulgaria, but has lived in the Netherlands for over 30 years, with a few intermezzos in England and Sweden. He explains what CogniGron is all about and why he is so enthusiastic about their work. ‘What’s so special about this work? The way that we combine principles from the fields of mathematics, artificial intelligence, and materials science, computer science and brain science. This approach is totally unique.’

Why do we need a new approach? Haven’t scientists already spent years working on a quantum computer that will solve all our problems?

‘They have. And a quantum computer will eventually compute data in a completely different way to that of the current generation of computers. But the problem is that quantum physics, the theory on which quantum computers are based, is still very much in the development phase. It needs a lot more fundamental work. So it will be a long time before the theory can be applied to practice, possibly another 20 years. We’re looking for a solution that will tide us over. A solution that can do things that even a quantum computer can’t do.’

What form will this solution take?

‘We are looking for a combination of concrete material properties and principles from brain science, that will allow us to perform certain computations. This will eventually result in a system that only responds to relevant signal changes from the surroundings, just like our brains. The brain doesn’t constantly process all the information it receives – that would drive us mad and we’d never get anything done. No, it’s not until we see something move, for instance, that a part of the brain starts to perform the corresponding, highly specific task of information processing. This is how brain-inspired computing will work. It’s much more efficient than our current computers.’

Which materials do you use?

‘They’re materials with specific combinations of, for example, electrical, magnetic, and thermal properties. We combine knowledge about this (and Groningen has been leading the field in this area for some time now) with knowledge from all the other disciplines. That’s what makes this work unique. Don’t get me wrong: we’re not trying to build artificial brains. We want to understand the basic principles of how the brain processes information and use it for efficient data processing. How is it possible, for example, that our brains are so fast, so powerful, so efficient, and have such a huge memory, yet use so little energy?’

But we only know a fraction of all there is to know about our brains… isn’t that frustrating?

‘Noooooo! That’s the whole point. That’s the challenge! To find something, to create something that provides a solution for a real societal problem, based on the workings of nature. Nature has got this sorted. Admittedly, it took millions of years, but it was thanks to certain basic principles. If we can manage to isolate these principles, we’ll have created something very special. That’s what makes this work so exciting.’

Is there a realistic chance that you’ll succeed?

‘As an engineer, I always try to break problems down into smaller, feasible projects. Projects that will make a difference – not just to academia, but also to society. I don’t think that we’ll find a generic solution, a complete, cognitive super-computer that can do everything. So we’ll focus on a particular function, or set of functions, and gradually unravel more and more basic principles. We should come close to a generic solution in the end. And in the meantime, we can fit all the tiny pieces of the puzzle into our current computers to make them more efficient too. This is what we’re doing now, which is great.’

Which applications are on the horizon?

‘Take things like weather and climate models. They are becoming increasingly dynamic and complex, which is why they are processing more and more data, with an increasingly high resolution in time and space. We are already reaching the limits. Currently, to get a really accurate forecast of tomorrow’s weather, a computer would need about a week to compute all the data it requires. As this would then be too late, it would be a pretty pointless exercise. It’s even more true of the worldwide climate models, which analyse larger areas for much longer periods. The current models simply aren’t that accurate.’

Could you give another example?

‘There are numerous applications in the medical world. Doctors want to use readings and measurements to offer patients more targeted, individual treatment. This too requires greater computing power. Take cancer patients receiving radiotherapy, for example. The doctor uses a scan to target the exact contours of the tumour, so that the radiation won’t hit the surrounding, healthy tissue. They use the same scan for a whole series of radiotherapy sessions, over several weeks. But we know that organs don’t remain static inside the body: they sometimes change their position and orientation. It would be much better to scan patients while they are being radiated. But this requires much more computing power than our current computers can manage.’

Do you really think that this will happen before the quantum computers are ready?

‘Yes, I’m 100% convinced. We can certainly produce the last example I gave within the next five to ten years. We’re already well on our way. I can’t stress enough just how much of a pioneer CogniGron is in Europe. There are very few initiatives that bring together the knowledge and capacities of all these disciplines so effectively. Groningen should be really proud of us.’

Text: Nienke Beintema, translation UVC (UG), illustration: Vera Post

CogniGron is led by the Scientific Director with support of the Coordinating Office for day-to-day operations.

The Scientific Director also chairs the Program Board, which oversees CogniGron’s research agenda, allocation of resources and the recruitment of new staff.

Financial planning decisions and the longer-term strategy are reviewed by the Supervisory Board. In addition, CogniGron has an international Scientific Advisory Panel to consult the project’s leadership on new strategic priorities for research.

Scientific Advisory Panel

Giacomo Indiveri, ETH Zurich, Switzerland

Julie Grollier, CNSR Thales, France

Heike Riel, IBM Zurich, Switzerland

Ivan Schuller, University of California, San Diego, USA

Rainer Waser, RWTH Aachen University & Peter Grünberg Institute, Julich, Germany

Yoeri van de Burgt, TU Eindhoven

Wilfred van der Wiel, TU Twente

Chris Eliasmith, University of Waterloo, Canada

Susan Stepney, University of York, UK

Supervisory Board

Hans Biemans, member of the Board of Directors, UG

Joost Frenken, dean Faculty Science and Engineering

Esther-Marije Klop, managing director Faculty Science and Engineering