A network to guide the future of computing

ScienceDaily (Sep. 13, 2012) — Moore's Law, the observation by Intel co-founder Gordon E. Moore that the number of transistors on a chip doubles approximately every two years, has been accurate for half a century. As a result, we now carry more processing power in the mobile phones in our pockets than could fit in a house-sized computer in the 1960s. But by around 2020 Moore's Law will start to reach its limits: the laws of physics will eventually pose a barrier to higher transistor density, but other factors such as heat, energy consumption and cost look set to slow the increase in performance even sooner.

At the same time, the world is in the midst of a data explosion, with humans and machines generating, storing, sharing and accessing ever increasing amounts of data in many different forms, on a multitude of different devices that require more energy-efficient, higher-performance processors.

How can computing systems, now facing a post-Moore era, meet this ever growing demand?

It is an open-ended question, but one that European researchers are working hard to answer, thanks in large measure to the efforts of HiPEAC (1), a 'Network of excellence' from academia and industry that has been helping to steer European computing systems research since 2004. Currently in its third incarnation, supported over four years by EUR 3.8 million in funding from the European Commission, the project has become the most visible and far-reaching computing systems network in Europe.

'HiPEAC was set up with three main goals: to bring together academia and industry, to bring together hardware and software developers and to create a real, visible computer systems community in Europe. On those fronts, and many others, we have undoubtedly succeeded,' says Koen De Bosschere, professor of the computer systems lab of Ghent University in Belgium and coordinator of the HiPEAC network.

HiPEAC's conferences and networking events are now attended by hundreds of academic researchers and industry representatives from Europe and beyond; the network's summer schools, workshops and exchange grants between universities are helping train researchers in new and emerging areas of computing systems theory and technology; and the project's biannual roadmap has become a guideline for both the public and private sector as to where research funding should be channelled.

'We now have a portfolio of between 30 and 40 computer systems projects that we are working with. The researchers involved come to our events, which have become one of the sector's main networking opportunities, and several projects have actually emerged from people meeting at our conferences,' Prof. De Bosschere notes.

He points, for example, to the EuroCloud project, which began in 2010 with the support of EUR 3.3 million in funding from the European Commission. Coordinated by microprocessor designer ARM in the United Kingdom, the project is developing on-chip servers using multiple ARM cores and integrating 3D DRAM with the aim of reducing energy consumption and costs at data centres by as much as 90 %.

The idea for the project first arose at the HiPEAC conference in Cyprus in 2009, Prof. De Bosschere notes. 'These kinds of networking opportunities are really showing their worth in spurring collaboration and innovation."

A roadmap of challenges and opportunities for Europe

Meanwhile, the HiPEAC Roadmap, a new edition of which is due to be published this year, has become something of a guidebook for the future of computing systems research in Europe.

'We didn't really set out doing it with that aim in mind, but the Commission took notice of it, consulted with industry on it, found the challenges we had identified to be accurate and started to use it to focus research funding,' the HiPEAC coordinator explains. 'Since we produced the first edition in 2008, EU funding in the sector has almost tripled and the next call will offer around EUR 70 million.'

For the short and medium term, the latest edition of the HiPEAC report concludes that specialising computing devices is the most promising but difficult path for dramatically improving the performance of future computing systems. In this light, HiPEAC has identified seven concrete research objectives -- from energy efficiency to system complexity and reliability -- related to the design and the exploitation of specialised heterogeneous systems. But in the longer term, the HiPEAC researchers say it will be critical to pursue research directions that break with classical systems, and their traditional hardware/software boundary, by investigating new devices and new computing paradigms, such as bio-inspired systems, stochastic computing and swarm computing.

'We can only go so far by following current trends and approaches, but in the long run we will nonetheless want and require more processing power that is more reliable, consumes less energy, produces less heat and can fit into smaller devices. More processing power means more applications and entirely new markets -- just look at what's happened with smartphones and tablet computers over the last five years,' Prof. De Bosschere says. 'For industry, it means that today, instead of a person having just one desktop or laptop computer, they may have three or four devices.'

And, in the future, he sees ever higher-performance devices doing much more than is possible or even imaginable today: bio-inspired neural networks powering data mining applications at 1 % of the energy consumption of today's data centres, for example, or smartphones that can analyse a blood sample, sequence the DNA and detect a virus in a few minutes, rather than the days it takes using laboratory computer systems at present.

'The potential applications for computing technology in almost every aspect of life are almost endless -- we just need to make sure we have the processing power to run them,' he says.

HiPEAC received research funding under the European Union's Seventh Framework Programme.

Share this story on Facebook, Twitter, and Google:

Other social bookmarking and sharing tools:

Story Source:

The above story is reprinted from materials provided by CORDIS Features, formerly ICT Results.

Note: Materials may be edited for content and length. For further information, please contact the source cited above.

Note: If no author is given, the source is cited instead.

Disclaimer: Views expressed in this article do not necessarily reflect those of ScienceDaily or its staff.

View the original article here

Search technology that can gauge opinion and predict the future

ScienceDaily (Aug. 16, 2012) — Inspired by a system for categorising books proposed by an Indian librarian more than 50 years ago, a team of EU-funded researchers have developed a new kind of internet search that takes into account factors such as opinion, bias, context, time and location. The new technology, which could soon be in use commercially, can display trends in public opinion about a topic, company or person over time -- and it can even be used to predict the future.

'Do a search for the word "climate" on Google or another search engine and what you will get back is basically a list of results featuring that word: there's no categorisation, no specific order, no context. Current search engines do not take into account the dimensions of diversity: factors such as when the information was published, if there is a bias toward one opinion or another inherent in the content and structure, who published it and when,' explains Fausto Giunchiglia, a professor of computer science at the University of Trento in Italy.

But can search technology be made to identify and embrace diversity? Can a search engine tell you, for example, how public opinion about climate change has changed over the last decade? Or how hot the weather will be a century from now, by aggregating current and past estimates from different sources?

It seems that it can, thanks to a pioneering combination of modern science and a decades-old classification method, brought together by European researchers in the LivingKnowledge (1) project. Supported by EUR 4.8 million in funding from the European Commission, the LivingKnowledge team, coordinated by Prof. Giunchiglia, adopted a multidisciplinary approach to developing new search technology, drawing on fields as diverse as computer science, social science, semiotics and library science.

Indeed, the so-called father of library science, Sirkali Ramamrita Ranganathan, an Indian librarian, served as a source of inspiration for the researchers. In the 1920s and 1930s, Ranganathan developed the first major analytico-synthetic, or faceted, classification system. Using this approach, objects -- books, in the case of Ranganathan; web and database content, in the case of the LivingKnowlege team -- are assigned multiple characteristics and attributes (facets), enabling the classification to be ordered in multiple ways, rather than in a single, predetermined, taxonomic order. Using the system, an article about the effects on agriculture of climate change written in Norway in 1990 might be classified as 'Geography; Climate; Climate change; Agriculture; Research; Norway; 1990.'

In order to understand the classification system better and implement it in search engine technology, the LivingKnowledge researchers turned to the Indian Statistical Institute, a project partner, which uses faceted classification on a daily basis.

'Using their knowledge we were able to turn Ranganathan's pseudo-algorithm into a computer algorithm and the computer scientists were able to use it to mine data from the web, extract its meaning and context, assign facets to it, and use these to structure the information based on the dimensions of diversity,' Prof. Giunchiglia says.

Researchers at the University of Pavia in Italy, another partner, drew on their expertise in extracting meaning from web content -- not just from text and multimedia content, but also from the way the information is structured and laid out -- in order to infer bias and opinions, adding another facet to the data.

'We are able to identify the bias of authors on a certain subject and whether their opinions are positive or negative,' the LivingKnowledge coordinator says. 'Facts are facts, but any information about an event, or on any subject, is often surrounded by opinions and bias.'

From libraries of the 1930s to space travel in 2034...

The technology was implemented in a testbed, now available as open source software, and used for trials based around two intriguing application scenarios.

Working with Austrian social research institute SORA, the team used the LivingKnowledge system to identify social trends and monitor public opinion in both quantitative and qualitative terms. Used for media content analysis, the system could help a company understand the impact of a new advertising campaign, showing how it has affected brand recognition over time and which social groups have been most receptive. Alternatively, a government might use the system to gauge public opinion about a new policy, or a politician could use it to respond in the most publicly acceptable way to a rival candidate's claims.

With Barcelona Media, a non-profit research foundation supported by Yahoo!, and with the Netherlands-based Internet Memory Foundation, the LivingKnowledge team looked not only at current and past trends, but extrapolated them and drew on forecasts extracted from existing data to try to predict the future. Their Future Predictor application is able to make searches based on questions such as 'What will oil prices be in 2050?' or 'How much will global temperatures rise over the next 100 years?' and find relevant information and forecasts from today's web. For example, a search for the year 2034 turns up 'space travel' as the most relevant topic indexed in today's news.

'More immediately, this application scenario provides functionality for detecting trends even before these trends become apparent in daily events -- based on integrated search and navigation capabilities for finding diverse, multi-dimensional information depending on content, bias and time,' Prof. Giunchiglia explains.

Several of the project partners have plans to implement the technology commercially, and the project coordinator intends to set up a non-profit foundation to build on the LivingKnowledge results at a time when demand for this sort of technology is only likely to increase.

As Prof. Giunchiglia points out, Google fundamentally changed the world by providing everyone with access to much of the world's information, but it did it for people: currently only humans can understand the meaning of all that data, so much so that information overload is a common problem. As we move into a 'big data' age in which information about everything and anything is available at the touch of a button, the meaning of that information needs to be understandable not just by humans but also by machines, so quantity must come combined with quality. The LivingKnowledge approach addresses that problem.

'When we started the project, no one was talking about big data. Now everyone is and there is increasing interest in this sort of technology,' Prof. Giunchiglia says. 'The future will be all about big data -- we can't say whether it will be good or bad, but it will certainly be different.'

Share this story on Facebook, Twitter, and Google:

Other social bookmarking and sharing tools:

Story Source:

The above story is reprinted from materials provided by CORDIS Features, formerly ICT Results.

Note: Materials may be edited for content and length. For further information, please contact the source cited above.

Note: If no author is given, the source is cited instead.

Disclaimer: Views expressed in this article do not necessarily reflect those of ScienceDaily or its staff.

View the original article here

New statistical model lets patient's past forecast future ailments

ScienceDaily (June 4, 2012) — Analyzing medical records from thousands of patients, statisticians have devised a statistical model for predicting what other medical problems a patient might encounter.

Like how Netflix recommends movies and TV shows or how Amazon.com suggests products to buy, the algorithm makes predictions based on what a patient has already experienced as well as the experiences of other patients showing a similar medical history.

"This provides physicians with insights on what might be coming next for a patient, based on experiences of other patients. It also gives a predication that is interpretable by patients," said Tyler McCormick, an assistant professor of statistics and sociology at the University of Washington.

The algorithm will be published in an upcoming issue of the journal Annals of Applied Statistics. McCormick's co-authors are Cynthia Rudin, Massachusetts Institute of Technology, and David Madigan, Columbia University.

McCormick said that this is one of the first times that this type of predictive algorithm has been used in a medical setting. What differentiates his model from others, he said, is that it shares information across patients who have similar health problems. This allows for better predictions when details of a patient's medical history are sparse.

For example, new patients might lack a lengthy file listing ailments and drug prescriptions compiled from previous doctor visits. The algorithm can compare the patient's current health complaints with other patients who have a more extensive medical record that includes similar symptoms and the timing of when they arise. Then the algorithm can point to what medical conditions might come next for the new patient.

"We're looking at each sequence of symptoms to try to predict the rest of the sequence for a different patient," McCormick said. If a patient has already had dyspepsia and epigastric pain, for instance, heartburn might be next.

The algorithm can also accommodate situations where it's statistically difficult to predict a less common condition. For instance, most patients do not experience strokes, and accordingly most models could not predict one because they only factor in an individual patient's medical history with a stroke. But McCormick's model mines medical histories of patients who went on to have a stroke and uses that analysis to make a stroke prediction.

The statisticians used medical records obtained from a multiyear clinical drug trial involving tens of thousands of patients aged 40 and older. The records included other demographic details, such as gender and ethnicity, as well as patients' histories of medical complaints and prescription medications.

They found that of the 1,800 medical conditions in the dataset, most of them -- 1,400 -- occurred fewer than 10 times. McCormick and his co-authors had to come up with a statistical way to not overlook those 1,400 conditions, while alerting patients who might actually experience those rarer conditions.

They came up with a statistical modeling technique that is grounded in Bayesian methods, the backbone of many predictive algorithms. McCormick and his co-authors call their approach the Hierarchical Association Rule Model and are working toward making it available to patients and doctors.

"We hope that this model will provide a more patient-centered approach to medical care and to improve patient experiences," McCormick said.

The work was funded by a Google Ph.D. fellowship awarded to McCormick and by the National Science Foundation.

Share this story on Facebook, Twitter, and Google:

Other social bookmarking and sharing tools:

Story Source:

The above story is reprinted from materials provided by University of Washington, via Newswise.

Note: Materials may be edited for content and length. For further information, please contact the source cited above.

Journal Reference:

Tyler H. McCormick, Cynthia Rudin and David Madigan. Bayesian Hierarchical Rule Modeling for Predicting Medical Conditions. Annals of Applied Statistics, 2012 [link]

Note: If no author is given, the source is cited instead.

Disclaimer: This article is not intended to provide medical advice, diagnosis or treatment. Views expressed here do not necessarily reflect those of ScienceDaily or its staff.

View the original article here

New statistical model lets patient's past forecast future ailments

ScienceDaily (June 4, 2012) — Analyzing medical records from thousands of patients, statisticians have devised a statistical model for predicting what other medical problems a patient might encounter.

Like how Netflix recommends movies and TV shows or how Amazon.com suggests products to buy, the algorithm makes predictions based on what a patient has already experienced as well as the experiences of other patients showing a similar medical history.

"This provides physicians with insights on what might be coming next for a patient, based on experiences of other patients. It also gives a predication that is interpretable by patients," said Tyler McCormick, an assistant professor of statistics and sociology at the University of Washington.

The algorithm will be published in an upcoming issue of the journal Annals of Applied Statistics. McCormick's co-authors are Cynthia Rudin, Massachusetts Institute of Technology, and David Madigan, Columbia University.

McCormick said that this is one of the first times that this type of predictive algorithm has been used in a medical setting. What differentiates his model from others, he said, is that it shares information across patients who have similar health problems. This allows for better predictions when details of a patient's medical history are sparse.

For example, new patients might lack a lengthy file listing ailments and drug prescriptions compiled from previous doctor visits. The algorithm can compare the patient's current health complaints with other patients who have a more extensive medical record that includes similar symptoms and the timing of when they arise. Then the algorithm can point to what medical conditions might come next for the new patient.

"We're looking at each sequence of symptoms to try to predict the rest of the sequence for a different patient," McCormick said. If a patient has already had dyspepsia and epigastric pain, for instance, heartburn might be next.

The algorithm can also accommodate situations where it's statistically difficult to predict a less common condition. For instance, most patients do not experience strokes, and accordingly most models could not predict one because they only factor in an individual patient's medical history with a stroke. But McCormick's model mines medical histories of patients who went on to have a stroke and uses that analysis to make a stroke prediction.

The statisticians used medical records obtained from a multiyear clinical drug trial involving tens of thousands of patients aged 40 and older. The records included other demographic details, such as gender and ethnicity, as well as patients' histories of medical complaints and prescription medications.

They found that of the 1,800 medical conditions in the dataset, most of them -- 1,400 -- occurred fewer than 10 times. McCormick and his co-authors had to come up with a statistical way to not overlook those 1,400 conditions, while alerting patients who might actually experience those rarer conditions.

They came up with a statistical modeling technique that is grounded in Bayesian methods, the backbone of many predictive algorithms. McCormick and his co-authors call their approach the Hierarchical Association Rule Model and are working toward making it available to patients and doctors.

"We hope that this model will provide a more patient-centered approach to medical care and to improve patient experiences," McCormick said.

The work was funded by a Google Ph.D. fellowship awarded to McCormick and by the National Science Foundation.

Share this story on Facebook, Twitter, and Google:

Other social bookmarking and sharing tools:

Story Source:

The above story is reprinted from materials provided by University of Washington, via Newswise.

Note: Materials may be edited for content and length. For further information, please contact the source cited above.

Journal Reference:

Tyler H. McCormick, Cynthia Rudin and David Madigan. Bayesian Hierarchical Rule Modeling for Predicting Medical Conditions. Annals of Applied Statistics, 2012 [link]

Note: If no author is given, the source is cited instead.

Disclaimer: This article is not intended to provide medical advice, diagnosis or treatment. Views expressed here do not necessarily reflect those of ScienceDaily or its staff.

View the original article here