Big brains, big data, and big opportunities

"The human brain has an average weight of 1.4kg and uses the caloric equivalent of just two bananas eaten per day. By contrast, a supercomputer of roughly equal performance would require a power station that could support half a city and would cover surface area equivalent to a few football fields." Image courtesy Richard Frackowiak.

With the 2^nd EUDAT conferencenow just one month away, iSGTW speaks in depth to Richard Frackowiak, who will be giving the event’s opening keynote speech. The conference is scheduled to take place in Rome, Italy, from 28-30 October and will address the challenges surrounding the creation of a collaborative, Europe-wide data infrastructure.

As well as being director of clinical neuroscience at the University of Lausanne in Switzerland, you’re also heavily involved in the Human Brain Project. Perhaps you could start by telling our audience a little about this…

The Human Brain Project is a large EU-flagship project, with the aim of harnessing the power of HPC to improve understanding of how the human brain is organized. We’re trying to use legacy data, stored in academic reports, journals, and databases, in order to apply big data analyses and build up a model — or better still, a theory — of how the brain is functionally and structurally organized. HPC, along with other advances made in IT in recent years, such as distributed databases, data addressing, and distributed analysis, make it easier to look for patterns in big data.

The project also consists of a neuroscience grouping and a medical informatics grouping, which aims to look for patterns in data from patients suffering from brain disease. The old, tried-and-tested medical technique of identifying syndromes and symptoms has reached its end and now needs to be replaced.

And are there things which the computer science community can learn from studying how the human brain works?

Today, computer scientists are seeing that Moore’s law is coming to a plateau — or at least reaching an asymptote — because of problems related to energy requirements. Computer scientists are always trying to engineer new computer architectures and one line of research involves building computers based on the human brain. There are even companies now producing so-called neuromorphic chips. They’re getting out of the domain of the digital, binary machines and are moving towards something more brain related. Consequently, these people need to have a theory of how the brain is organized and works.

The human brain has an average weight of 1.4kg and uses the caloric equivalent of just two bananas eaten per day. By contrast, a supercomputer of roughly equal performance would require a power station that could support half a city and would cover surface area equivalent to a few football fields. Given that both brains and supercomputers are made up of matter, there’s clearly some fundamental correspondence between the two. So, the real challenge for computer scientists now is to see what exactly this correspondence is. For that, they need the neuroscientists.

Equally, the neuroscientists need the computer scientists to help them do their work. They really need a model of the brain, so that they can use the masses of data that’s been compiled in the field of neuroscience over the last 30-40 years to gain a holistic view of how higher functions arise from organic matter in the human brain.

How does the work done through the EUDAT project help researchers like you?

EUDAT is a particularly interesting project, because it has the potential to facilitate and help with a lot of the things which we’re going to try to do with the Human Brain Project. Of course, there are also other projects in other fields which will also benefit from the sort of harmonization that can be brought about by projects like EUDAT.

With all of these projects, there are, of course, also certain pitfalls which need to be avoided. They’re generally aware of these pitfalls, but they need to keep them in mind at all times. For example: over the last 10 years, information technology has progressed at such a fantastic pace — it’s really the Usain Bolt of research. Yet, it is exactly because of this, that we need to always be aware that what we do today may not actually have any relevance at all to what we do tomorrow. When we started thinking about the Human Brain Project three years ago, we were all fascinated by the prospects of cloud computing. However, over the last three years, it’s become clear to us that distributed databases are actually the best way for us to go forward. Researchers are often very jealous and paranoid people, who want to hang on to their data and don’t want to move it around too much. More validly, hospitals may not be able to move data around because of corruption problems; they don’t want data to be copied for all sorts of privacy reasons.

Projects like EUDAT make the transfer of knowledge about what is possible and what will be possible in the near future, in terms of computing and data management, a lot quicker. This enables scientists and clinicians to be a lot more adaptable in their thinking about what they can potentially do to tackle the problems they are researching. As such, I think these projects are extremely useful.

A big part of the EUDAT project is about bringing data together from different fields, such as linguistics, climate modelling, biomedical science, etc. How useful are these efforts for researchers like you, given the diversity of the fields from which the data is drawn?

You can look at data in various ways: you can classify data according to subject, according to mathematical type, information type, etc. This is an area full of complexity, but the people working on projects like EUDAT, for instance, aren’t simply thinking in terms of just bringing as much data together as possible. Instead, they’re thinking about which data for whom, how best to share it, and what do they want to do with it. It’s all about making the right data available to the right people in the right format for answering specific questions.

Finally, could you tell our audience what talks or workshops you’re most looking forward to attending at the upcoming EUDAT conference?

I’m looking forward to finding out more about EUDAT’s work in general. However, I’m also interested in the talk by Ewan Birney from the European Bioinformatics institute on annotating the human genome. Finally, I think that the Wednesday morning session, Towards Global Data Infrastructure Components, should provide a great opportunity to see how research in other areas is progressing and to get an idea of what the future is going to look like in terms of data.