With SFI preparing this week to channel more millions towards high-end RD projects across the nation's third-level institutions, have previous investments really lived up to expectations, asks JJ WORRALL
THE ANNUAL figures usually catch the eye – €44 million in 2011, €25 million the year before – and in a few weeks, Science Foundation Ireland (SFI) once again announces just how much money will be granted under its Principal Investigator (PI) programme.
Supporting research and development in various forms of software, hardware, telecommunications, biotechnology and sustainable energy, PI grants range from €100,000 to €500,000 per year for three- to five-year projects. Far from being cocooned within academia though, they attract the support of global IT leaders such as IBM, Microsoft, EMC, Disney, Intel and Cisco as well as local names like Storyful and Havok.
“In the current climate it’s important that you can demonstrate that the research has a commercial element,” admits UCD’s Prof Pádraig Cunningham who gained a PI programme grant of more than €450,000 in 2006 for a three-year data-mining project.
Like others who have received SFI funding in recent years, he tells how there’s been “a certain shift” in SFI’s attitude towards research funding of late, shifting from “basic research” towards “industry-focused research”. This “value for money” approach reflects the type of pressure most government bodies face these days, though it does beg the question whether the millions spent previously in the last decade truly fulfilled their potential?
The Irish Times spoke to three people whose PI projects have come to the end of their life cycle to find out.
PROJECT TITLE: Employing Artificial Intelligence to Make Constraint Programming Easier to Use for Decision Making
Investment: €3,338,688
Details: With more than €3 million in SFI funding, this four-year project was the investment which really kick-started the acclaimed Cork Constraint Computation Centre (or 4C as it’s also known), in UCC. Professors Barry O’Sullivan and Eugene Freuder intended for the funds to be focused on “making constraint programming easier to use by using artificial intelligence techniques”.
Freuder once said 4C’s work can be described as helping computers to make “better decisions. Using artificial intelligence O’Sullivan and Freuder aimed to automate this process by “teaching” computers how to solve various problems.
While there were no initial commercial partners involved, their work over the four years included contributions from Microsoft, Xerox, BT and Intel. While a $100,000 Google Research Award was bestowed upon O’Sullivan and Frueder’s team to help the development of an advanced event scheduler.
Value for money? The PI program grant resulted directly in what O’Sullivan believes was “the highest rate of technology transfer of any similar projects”. Two start-up companies – Keelvar, which creates optimised web-based market systems for financial traders, and location analytics business ThinkSmart – grew from the work done under O’Sullivan and Freuder’s watch. In addition, about 50 4C staff had their work funded or co-funded by the grant.
Indeed, O’Sullivan and his 4C colleagues have also helped bring commercial investment towards the Rebel County. “We’ve helped the IDA bring companies to Ireland like United Technologies and Quest Software,” says O’Sullivan.
Meanwhile EMC’s Research Europe lab – which is focused on cloud computing solutions, data centre optimisation and data centre energy management – is actually co-located on the UCC campus in partnership with 4C.
What’s next? “During those four years, we had those two start-ups, we registered about two dozen inventions, we had about eight intellectual property licences and two patent applications arise from the research we did as well,“ says O’Sullivan. “That’s the path we’re continuing down.”
PROJECT TITLE: Metropolis: Supercrowds for Multisensory Urban Simulations
Investment: €1,893,842
Details: The scope of this five-year project was to create more realistic virtual crowds. Aimed for use across movies, video games and within neuroscience research, Prof Carol O’Sullivan was at the forefront of the project which attracted the attention of Disney, Havok, Natural Motion, IBM and Intel, with some partners donating “hundreds of thousands” of euro in terms of software and hardware.
O’Sullivan worked alongside fellow principal investigators Dr Steve Collins and Professors Fiona Newell and Henry Rice, with about 20 post-graduate and post-doctoral students employed as researchers as well. Collins’ spin-off company, Kore Virtual Machines was one of the main successes of Metropolis; eventually being bought up by games engine Havok for an undisclosed amount.
The goal of creating more realistic crowds has been achieved, according to O’Sullivan, who says that “when industry people think of crowd simulation or crowd graphics they think of Dublin as being one of the major players, if not the major player due to Metropolis”.
Value for money? SFI and others certainly seem to think so. That’s why O’Sullivan, Newell and others have been asked to work on two other major new projects. First there’s Captavatar, which sees the group of researchers team up with Disney for an inter-disciplinary project combining computer graphics and social cognitive neuroscience, to create virtual humans with “maximum social appeal”.
Also, O’Sullivan and her colleagues are working on Verve, a European-wide project coordinated from TCD which will attempt to use the virtual crowds created during Metropolis to help in treating the elderly and those with neurological disorders.
Funded by the European Commission Seventh Framework, O’Sullivan explains that the work involves using virtual simulations of familiar environments. This is to help those suffering from neurological disorders to gain a “personalised virtual reality experience” which can be used to ease stress or even maintain cognitive functions.
“You could take a couple of photos of an elderly person’s street, recreate the environment and then put our ‘crowds’ from Metropolis in this scene,” she says.
What’s next? Alongside continuing the Verve and Captavatar projects, the researchers involved in Metropolis are preparing to work on a project with RTÉ to create a “digital resource” of James Joyce’s Dublin. Alongside Dublin-based motion graphics experts Noho, they hope to create realistic and interactive versions of various Joyce tales.
Meanwhile, O’Sullivan reveals that the IP behind the virtual crowds created during Metropolis are being adapted by the German Research Centre for Artificial Intelligence for use in the XML3D, which will be the format of the 3D World Wide Web, the development of which she says is “a lot closer than people think”.
PROJECT TITLE: Data Mining Between Unsupervised and Supervised Techniques
Investment: €446,402
Details: Interest in the data mining field has risen immeasurably since Prof Pádraig Cunningham, Dr Derek Greene and their team at UCD embarked upon this three-year project in 2006. Well, then again, maybe they could measure it considering the intellectual property created under their watch.
Once perhaps just an academic concern, the analysis of large clusters of data has become an obsession for social network behemoths such as Facebook and Twitter in particular.
The grant figure involved in Cunningham and Greene’s project may be dwarfed by other PI programme beneficiaries, but having originally being intended as a predominantly theoretical research project, it turned into an attention-grabbing piece of work that got practical results in several industries. For instance, electronic fault diagnosis saw the team focusing on how to predict when electronic devices such as wifi masts are coming towards the end of their life.
Then came the creation of a data-mining and text-clustering product for Polecat, a market intelligence provider with offices in Dublin and the UK. With the company gathering information from 300,000 articles and four million blog posts per day, they use software algorithms licensed to them by Cunningham and Greene’s group of researchers to decipher results related to their clients such as Shell, BT and Microsoft. Work on enterprise-level social media products with Cisco was another element of Cunningham and Greene’s project.
In a slightly more left-field strand of research, biological networks were also explored. “In bio-science the old idea that there’s a gene which causes cancer is gone. and now we realise that cancer, and other disorders, are caused by networks of genes,” explains Cunningham.
The group’s work then focused on how to decipher data from these networks to help understand the workings of the disease in minute detail.
Value for money? Considering the relatively small investment the first time round, this could have flown under the radar. Instead, it turned into a pilot project for data mining which could have huge commercial possibilities. To explore this further Cunningham now heads up Clique, a large research cluster based in UCD.
Focusing on social network and bio-network analysis and employing about 30 researchers, it’s built on a €3.56 million grant from SFI, while funding from partners such as Idiro Technologies, DataHug, Storyful and Accenture has pushed the budget for the research cluster beyond €5 million.
What’s next? Clique is helping build more sophisticated data clustering tools for their commercial partners and licensing out software to various SMEs.
