Computer Science

Computers are continuing to change our lives as the sophistication of the human machine interface develops and becomes more powerful

Backed up with software that learns from its mistakes, artificial intelligence is already augmenting human performance and intelligence in ways that would have been impossible only a few years ago.

Efforts to develop a human-like synthetic intelligence have advanced the growing supply of the feedstock needed to make it work, vast amounts of data that can be mined for hidden patterns. The computer programme ploughs through the mass of data but it is still up to the human operator to pluck out what is important.

UCD Professor Barry Smyth lives in this science fiction-like world, developing systems that support human-centred services that can really change lives and the way we currently do things. Artificial intelligence will change lives, he says, with better medical diagnostics, cars that drive themselves and a new kind of “personal assistant” that can do anything from keeping you connected to the office to helping you run a faster marathon.

“Artificial intelligence is not so much a replacement as an augmentation,” he says. It is not about replacing humans with robots. “It will be more interesting in that we will work together with machines and then real power emerges.”

Professor Smyth holds the Digital Chair of computer science in University College Dublin’s College of Science. He is also a founding director of the Insight Centre for Data Analytics, a multi disciplinary, multi institutional research centre that supports 450 researchers working in big data, artificial intelligence and machine learning. It collaborates with 80 partner companies and has a budget of about €100m supported by Science Foundation Ireland.

“Artificial intelligence is not so much a replacement as an augmentation. It is not about replacing humans with robots.”

“It is about machine learning. It discovers patterns in the data – but the challenge is to get actionable insights.”

Today we all live under an avalanche of data that pours in from all directions. The question is how to make best use of this raw data to improve how we live, work, and play. Smyth can see how artificial intelligence can contribute by working with rather than against humans.

“A key technology is machine learning,” he explains. “which is a way of training machines to discover subtle patterns in vast quantities of data. The trick is helping machines to identify interesting patterns, actionable patterns that suggest something to do.”

This reveals the partnership possible with data mining, with the computer doing the mining while the human provides the refining. The two can accomplish more by working in tandem, than either can on their own, to solve bigger problems and deliver better improvements.

Examples of this can be seen in companies such as Amazon or online travel companies who make use of “recommender systems”. If the system knows you have purchased say the first series of Game of Thrones then you might respond to an online recommendation that you purchase series two or perhaps the whole box set.

Smyth sees how the system could be made “smarter” by analysing the words used in reviews sent by contributors to say Natural language processing could search for words that convey positive sentiment such as “good” or “excellent” or negative views like “bad”.

“Now you have a different way to compare hotels liked and disliked by similar people,” he says. “Some of the excitement about machine learning is you can start to take advantage of this kind of data at scale.”

Even more sophistication could emerge if the computer algorithm was required to explain the recommendation it has made. If the explanation is convincing then “there is greater likelihood that a person will respond to that suggestion”, he argues.

Smyth is working on recommender systems but is also involved in a project linked to marathon running and exercise more generally. We already have smart phones and smart watches that measure things such as heart rate, blood oxygenation and running pace, but the research would deliver a much higher level of sophistication.

Smyth is a marathon runner and wanted to know whether endurance running performance could be improved if a smart system was there to help. For example all runners are warned not to rush off at the start but there was no data to prove this made any real difference.

In fact it turned out to be very important, particularly for recreational runners, he said. “What you do at the start of a race has a great impact on how you finish.”

The use of artificial intelligence and machine learning was brought to bear on the problem and then became part of a new kind of “digital personal assistant” for personal fitness and training that could help the marathon runner deliver a peak performance. It is able to provide kilometre by kilometre advice on pace achieved and target pace. It takes account of the actual topography of the running route and even factors in the weather forecast if it is likely to have an impact. “It’s not difficult to imagine how, in the near future, similar ideas could be extended by taking advantage of our eating habits and sleep patterns to provide a more complete training picture,” he suggests.

He notes however that a large component of the big data mountain is actually personal information and in many cases actual medical data captured by fitness devices. Controls are needed to limit who can see or access this information. In the coming years it is likely that individuals will have much more control over their personal data, he says.

“It will change the way the internet works in the future. In the past there was open data but in the future (third parties) might have to pay for services if we don’t want to surrender personal data. People will come to realise how valuable their personal information is.”

Where will this technology take us in the coming years? “These technologies come into their own when we learn to work with them rather than as (human) replacements,” says Smyth who admits to being a technological “optimist”. If we use them in a responsible way the changes they can bring are to the good of everyone.”

“These technologies come into their own when we learn to work with them rather than as (human) replacements.”

Professor Barry Smyth,
UCD School of Computer Science
Science Centre, North Belfield
Dublin 4, Ireland
T: +353 01 716 2473

Featured Research Papers

Professor Barry Smyth

Professor Barry Smyth holds the Digital Chair of Computer Science in University College Dublin and is a founding director of the Insight Centre for Data Analytics. He is a Fellow of the European Coordinating Committee on Artificial Intelligence (ECCAI) since 2003 and a Member of the Royal Irish Academy since 2011. In 2014 Barry was awarded an Honorary Doctor of Technology (Hons. D.Tech) from Robert Gordon University in the UK. Barry was the Director of the Clarity Centre for Sensor Web Technologies (2008 – 2013) and has previously held the position of Head of School for the School of Computer Science and Informatics in UCD.

Barry’s research interests fall within the field of Artificial Intelligence and include case-based reasoning, machine learning, recommender systems, user modeling and personalization. Since 1992 he has published over 400 peer-reviewed papers. Barry’s research has attracted more than 13,000 citations. He has a h-index of 58 and he has received more than 20 best paper awards for his work. In 2014 he was named the SFI Researcher of the Year.

Barry’s research interests extend beyond the laboratory and over years he has established a track-record for successfully translating his research into commercial opportunities and received numerous awards for his commercialisation endeavours. In 1999 Barry co-founded ChangingWorlds based on his personalization research and helped grow the company to more than 150 employees before it was acquired by Amdocs Inc 2008. In 2008 he co-founded HeyStaks based on his social search research. He has helped HeyStaks to raise more than €3.5m in venture capital funding to date and continues to advise the company on its technology and market strategy. Barry is actively involved in the Irish startup scene as an advisor and investor and he serves on the boards of a number of local startups. He is also a member of the Irish Times Trust.