Annual Review 2014/15

Delivering big data solutions

University researchers are part of a UK-wide initiative that seeks to draw meaning from the explosion of digital output and realise its economic potential. 

Vice-Principal High Performance Computing, Professor Richard Kenway, at the ARCHER facility, housed at Edinburgh’s Technopole.
Vice-Principal High Performance Computing, Professor Richard Kenway, at the ARCHER facility, housed at Edinburgh’s Technopole.

If the future is to be shaped by the dizzying amounts of data people produce, then history suggests Edinburgh will be uniquely placed to harness the benefits. For 50 years the institution has led many advances that have shaped the computer age. Now its researchers are part of a UK-wide initiative that seeks to draw meaning from the explosion of digital output and realise its economic potential. 

The new venture, named after the computer pioneer and code-breaker Alan Turing, focuses on ways of collecting, organising and interpreting large sets of digital information – commonly known as big data. Based at the British Library, the Alan Turing Institute will draw on Edinburgh’s expertise in computational, mathematical and social sciences. 

In 2015 it was announced that the Institute will be headed by Edinburgh alumnus and former computer science researcher at Edinburgh Professor Andrew Blake. Its aim is to make breakthroughs that will produce new algorithms – step-by-step sets of operations – that are needed to address real-world problems. 

The Institute is backed by £42 million of UK government money and £10 million from Lloyd’s Register Foundation. Each university involved – Cambridge, Edinburgh, Oxford, UCL and Warwick – will contribute £5 million. Government Communications Headquarters (GCHQ) and Intel Corporation have also announced their intention to become partners. With backing from the Engineering and Physical Sciences Research Council (EPSRC) it will develop strategic links with industry and commerce, seek to improve cyber security and train the next generation of data scientists. 

Demand for analysis 

The idea that universities should be drivers of economic growth as well as game-changing research is a familiar one, but the digital revolution takes these aspirations to new heights. Google Chairman Eric Schmidt put it eloquently in 2010 when he estimated that between the dawn of civilisation and 2003, humankind had created five billion billion bytes of information – a total that is now produced every two days. With this flood of data – most of it unstructured, much of it incomplete and some of it wrong – comes a demand for meaningful analysis and heightened expectations from companies. 

The University’s Professor Richard Kenway, Non-Executive Director of the Institute and Vice-Principal High Performance Computing, urges caution. “Data science has a vast amount of promise, but we mustn’t get carried away and take at face value what data appears to be telling us,” he says. 

“We need a deeper understanding of the underlying mechanisms and, from this, new algorithms. Data only describes what has already taken place – we can’t guarantee that the future is going to resemble the past.” 

This mission to produce insights, services and products is as far-reaching as the data itself, but Edinburgh scientists have faced similar challenges before. Artificial intelligence pioneer Professor Donald Michie – a wartime colleague of Alan Turing at Bletchley Park – assembled his fledgling research group at Edinburgh in the 1960s. 

Before long, this remarkable polymath – with degrees in anatomy, genetics and biological sciences – was helping to bring about the world of robots, computer games and search engines. Professor Michie’s multi-disciplinary approach and willingness to engage with business – to say nothing of his visionary genius – would sit well in the Institute that bears his friend’s name. 

By the time computers had become an indispensable research tool in the 1980s, Edinburgh was ready to up the stakes. An emerging generation of physicists – among them Professor Kenway – was ready to exploit a new type of computation that was creating previously unthinkable opportunities. Parallel computing – which enables many calculations to be completed simultaneously on different microprocessors – was precisely what the worlds of industry and commerce were waiting for. 

We are creating the new methodology of data science.

Professor Richard Kenway

In the next decade, as the World Wide Web began its spectacular ascendancy, Edinburgh was again in the vanguard, becoming a key player in the emerging field of e-science. Scattered networks of scientists, working across continents and different disciplines, used a turbo-charged computational network called the Grid to turn shared data into knowledge. Once more, business liked what it saw. 

Computing leadership

Standing still was not an option. Edinburgh became the home of the UK’s first super computer HECTOR in 2007, and this honour was re-established with the placing of its successor, ARCHER, in the hands of the University’s EPCC researchers in 2014. Capable of more than one million billion calculations a second, the £43 million ARCHER system provides high performance computing support for a range of research and industry projects and is number one in the UK, and 40th in the world, in terms of raw computational performance. 

Among the Alan Turing Institute’s earliest projects is a collaboration with supercomputer manufacturer Cray and the EPSRC to upgrade ARCHER’s analytics capability. 

The Institute’s joint programme with Lloyd’s Register Foundation will develop data-centric applications in engineering to enhance safety at sea, on land and in the air. 

Plans are also under way to work with GCHQ to develop data-analysis methods that can be applied in open access and commercial environments. 

For Professor Kenway, such projects demonstrate the breadth of the challenges ahead. “We are creating the new methodology of data science,” he says. 

“The data scientist is more akin to an Olympic pentathlete than an elite runner who focuses on a single event. He or she – like Donald Michie – will have to excel in a range of disciplines.”