Q. What is ICES?
Robert Bishop: The acronym “ICES” stands for the International Center for Earth Simulation. We see this as a long-term issue. It’s a very ambitious mission to bring together all the sciences which are currently siloed separately. They don’t talk to each other enough.
As a consequence I don’t think the destiny and the future of the planet is well understood because, until we integrate these sciences horizontally, and have them talk to each other, we really don’t know the best model to simulate the future of the planet.
Q. How does this compare to Google Earth?
RB: Google Earth is a beautiful fly-through of a static database. We want to put dynamics on top of that database so that we could see the weather and the climate, for example, and the environment and biodiversity and even the interaction of the earth with the sun through space weather and all the layers of the ocean and the layers of the atmosphere: We’d like to see it dynamically in a model.
It’s a very complex model, of course. It takes a lot of computing power to create that model and that visual output. But Google Earth is a very good first step. The public is very happy with it as a way of displaying static data. We want to create simulated model dynamic data.
Q. What is the biggest challenge?
RB: Today we have petaflop computing. The top ten computers of the world are all above one petaflop. The next level up in computing has a thousand times more power, called an exaflop. That’s where we’re heading right now in the industry as a whole.
I believe it will take an exaflop to solve the problem that ICES is presented with, which is the integration of all the sciences. There’s too much complexity to compute with today’s level of computing. We can’t get enough compute power to do the job.
We are expecting a thousand times more power in the next ten years and maybe a million times more power in the next 20 years. We are expecting that and we will need it to solve the problems of ICES.
Q. If this all works out, will we then have the power to predict natural disasters?
RB: Well, I think in the case of the natural sciences, beginning with the ocean, the weather and the climate, those things are becoming more and more predictable. If you take the history of the past thirty years, we have improved numerical weather prediction dramatically. I would say today it’s 90% predictable out to five days. Out to ten days its maybe 80% predictable.
The question is, as we go to out a month, a season, a year, interannual, and then ten years or more: what level of predictability can be get if we have more compute power and more data to assimilate.
I think the answer to every one of those questions is that we will be improving predictability but there will never be 100% predictability.
So, in talking to the public, the question is how we talk about uncertainty. How we engage the public in a broader discussion about the future of the planet on the one hand, but on the other hand we have the caveat of how much uncertainty is there in that discussion.
Q. What about citizen science tools, such as World Community Grid?
RB: The citizen science movement is very powerful. I like it, and I think it’s getting more powerful, especially as smart cellphones become available to perhaps the entire population of the world.
The idea of having 7 billion smartphones available for computation is pretty good. It requires a lot of software to bring it together, but I’m thinking more about the data that can be provided on those phones.
For example, over the next 10 years I think it is possible that these phones contain a single-chip automatic weather station. Together with GPS, that data would be very handy for a global simulator to simulate the weather in the future. I look upon it as a data entry point.
Today, if you take some of the developing countries, which have not had many computing resources of their own, the smartphone is actually penetrating those countries very quickly. They could benefit from having output from a global simulator, which would tell them what’s going to happen in their local weather.
It would be a regional downscaling from a global model, which would be helpful for running their own economies.
Q. What can 425,000 IBMers do to help?
RB: First of all, it’s a fantastic group of people, a fantastic company. They’re very professional. I’ve known this company for 45 years of my professional career.
I would love to see a partnership develop with IBM and I think IBM’s concept of standing for a smarter planet—using more intelligence. The ICES’ mission fits exactly within that scope.
The employees of IBM are very powerful and highly respected around the world. So having them contribute their ideas, their engagement with ICES would be wonderful. I would suggest they take a look at our website and let me know how you’d like to connect and participate. I’m quite open.
This is a not-for-profit mission. It’s Swiss-based, a neutral country with a lot of scientific intelligence and integrity. From that point of view I think it could be a good focal point for such a global project. I would very much open up to any inputs or suggested methods of collaboration with IBM and its employees.
Thank you for your time this evening.
RB: It’s been a great pleasure to be here.
A copy of her presentation can be downloaded here.
Silva is currently working as an assistant professor at Radboud University Nijmegen and as a research visitor at CWI in Amsterdam, the Netherlands.
Established in 1990 the IBM Scientific Prize is the most important Portuguese award for high scientific merits in computer science and is meant for researchers younger than 36 years. This year also marks the first time a woman has been recognized.
Alexandra Silva finished a degree in Mathematics and Computer Science at the University of Minho (UMinho). Being a PhD student at the Centrum Wiskunde & Informatica (CWI) in the Foundations of Software Engineering (SEN3) research group, she defended her thesis ‘Kleene Cooalgebra’ at the Radboud University Nijmegen (RUN) in 2010, passing ‘cum laude’. She was awarded a postdoctoral fellowship in the HASLab group at UMinho and also won a contest to become Assistant Professor in Nijmegen. In the past, she was also research visitor at Cornell University, and post-doctoral researcher at CWI.
We started the Genographic Project with National Geographic six years ago, and the first thing we worried about was how the general public would respond to our request for their DNA. There’s nothing more personal than that.
Well, we ran out of our initial supply of 30,000 kits – and reached 100,000 DNA samples – in the first year! (I personally didn’t think we would hand out the 30,000 kits over the five years of the project.)
What did we do right?
Having a thorough and well-communicated ethics policy led to why we succeeded in collecting the 470,000 samples. We assured the public that though we’re taking their DNA, participation would be anonymous; the DNA would be used only for determining the migratory history of humankind; and that their DNA would not be analyzed for clinically informative markers (such as a family history of hypertension).
Tracking paternal and maternal ancestry
At the outset, we analyzed two pieces of genetic evidence in the Genographic Project. First, looking at the entire human genome, certain fragments pass from one parent to a child that does not mix with the genetic information from the other parent. In males, this is the Y chromosome.
The Y chromosome goes from father to son with almost no modification. But the transition is a bit like copying a book by reading and re-writing – occasionally there’s a typo. That typo is what we call a mutation, or a marker of descent.
A mutation may only appear in one copy of one instance of DNA that could be passed from father to son. For example, only one of two brothers may get the mutation. Now, they are marked by this difference and the brother with the mutation (and his descendants) will carry that marker. We can see that mutation and track the male descendants for generations, back to the first male who showed that marker.
The other piece of genetic evidence we tracked also comes from genetic fragments that can only pass from one parent to a child. DNA contained in mitochondria, structures in the body of every human cell, is passed from a mother to all of her children and provides a means of tracing a maternal line of ancestry.
Y chromosome and mitochondrial DNA constitute less than 1 percent of the human genome. The rest of the genome is not directly inherited from a single parent. Rather, it undergoes a process of recombination, effectively shuffling fragments of DNA from each parent to create the unique genome of each child.
Tracing the ancestral history on genomic regions besides Y and mitochondrial is a daunting task, confounded by the active recombination that occurs every generation.
Crunching the genetic data
|IBM in Genetics|
IBM has invested in genomics and computational biology, for more than 15 years. These are disciplines that inform the life science, pharmaceutical and biotechnology industries. IT plays a vital role in enabling new science and discovery in biology, transforming the field into an information science.
We worked with population geneticists in various regional centers across the world to analyze this genetic data in populations of Sub-Saharan Africa, North Africa, the Middle East – and other regions are being concluded. Results published so far about the migratory history of the earliest humans in Africa include genetic evidence of relatively recent migratory events, such as the arrival of Crusaders in the Middle East, and the spread of the Phoenicians into the Mediterranean.
But the computational task of analyzing the data of our 470,000 samples is not a brute force exercise. A supercomputer is not required. Laxmi Parida, a member of the IBM team led a three-year effort in collaboration with Jaume Bertranpetit at the University of Pompeu Fabra in Spain, to develop an elegant algorithm that reconstructs recombinant history of the genome – using only workstations. They analyzed markers on the X chromosome of 1,240 male participants, from 30 different ethnicities across Africa, Middle East, Europe, and Asia.
On the point of migration, our findings showed that Eurasian groups were more similar to populations from southern India, than they were to those in Africa. This supports a southern route of migration from Africa via the Bab-el-Mandeb Strait in Arabia, before any movement heading north. It suggests a special role for South Asia in the “out of Africa” expansion of modern humans.
Keep in mind, exactly which direction or route humans took in migrating out of Africa is still not settled. This new genetic evidence suggests that other fields of research such as archaeology and anthropology should look for additional evidence on the migration route of early humans to further explore this theory.
How to participate
|Photo Credit: ETH Zurich/ Oliver Bartenschlager|
- Jim Schatz, Department Head, Johns Hopkins Applied Information Sciences
- David Aronoff, Flybridge Venture Capital
- Erik Kruse, Marketing Strategist, Ericsson
- Dr. Ashifi Gogo, CEO, Sproxil
"Projecting to exascale, we're heading toward 10-to-the-20 [operations per second]. The universe does 10-to-the-120 ops! There isn't a word for that. There's 10-to-the-100 head room we haven't reached. Why not?"
"If you look at [the health care industry], it's not that there's a lack of data. But if you look across the silos of the industry, whether it's pharmaceuticals, or medical devices, or medical providers and patients, it's a bunch of independent pools of information. So how do we bring it together ... and determine what's relevant?
We believe that [IBM Watson] can help bring all involved into the conversation, and help WellPoint share the data."
Dr. Steven E. Koonin, the undersecretary for Science at the US Department of Energy, spoke about how computing systems are becoming more energy efficient.
“There are three challenges associated with energy. First is energy security … [in part because] the fact that most of the world’s conventional oil reserves are not accessible to the big independent oil companies, but are rather in the hands of the national oil companies.
“Second is the U.S.'s competitiveness. Even with technologies invented and refined in the U.S., it is not manufacturing the new energy technologies.
“And third is environmental impact … most-famously carbon dioxide and its effects on climate. But equally important are water issues. Half of the water withdrawals in the U.S. are due to generating electrical power – more than in agriculture.”
“What we’re talking about is the modern age of genetic engineering, where we can imagine living foundries – that is, synthetic biology for synthetic chemistry, making a wide range of products that are perhaps known in nature, but not necessarily made by the organisms that we would like to make them in the most-efficient way [such as pharmaceuticals and fuels].”
As part of IBM’s Celebration of Service during the company's Centennial year, IBM Research – Tokyo employees are helping the Japan Braille Library with this project. The research team is also developing an electronic book improvement system for the Japan Braille Library. It allows volunteers to collaboratively correct Optical Character Recognition (OCR)-scanned characters via a web browser.
Making Braille books
In Japan, volunteers are the driving force in making accessible books. The process to create Braille books largely relies on manual input, using a standard computer keyboard -- or creating voice recordings for audio books. It can take several months to make an accessible book.
The Japan Braille Library receives requests from library users about which new books to make available. From the wish list, they scan some of the books for the project. The collaborative character correction interface displays a group of OCR-scanned characters recognized as the same character – allowing IBM volunteers to make a batch correction.
Volunteers also use the collaborative ruby correction interface (a small Japanese syllabary character reading aid printed right next to a kanji character). After corrections are made, the digitized books are converted into the international standard-based format and added to the Sapie Library collection.
"Since the kanji characters with ruby tend to be difficult, I enjoyed it because it was like challenging the kanji kentei (Japanese kanji aptitude test)," said an IBM Research volunteer.
IBM Research and Braille technology
In 1986, IBM researchers in Tokyo developed Braille word processing software that significantly sped up Braille translation work. Two years later, IBM collaborated with Braille libraries and volunteer groups across Japan to create an inter-library Braille network. It allowed visually impaired persons to access Braille books from home. The Braille network was migrated to the Internet by the Ministry of Health, Labor and Welfare in 2000. Today, it is Japan's largest bibliographic database, used by visually impaired citizens nationwide.
David Ferrucci, IBM Fellow and principal investigator on the IBM Watson Project, discussed what is on the horizon for the fields of artificial intelligence, natural language processing and machine learning at the recent “Frontiers of IT” IBM Research Colloquium.
“I think the impact from robotics on a new age of information science, and even on programming the genome, are all areas that are going to have a big impact in the near future,” Dr. Ferrucci said.
Still an early prototype, the technology converts the shape of an object, such as an image shown on a webpage, into a stream of sound. By listening to the sounds, the visually impaired can mentally feel the shape or better-understand turn-by-turn directions. The technology may also be useful as a real-time guide for online maps which shows route to destination via a mobile phone.
According to a preliminary experiment conducted by the IBM researchers, both blind and sighted people quickly adapted to following directional sound, and recognizing shapes such as randomly drawn paths, letters in the alphabet, and numbers. People with musical training tended to adopt the technology at a faster rate. And some of the participants felt as if they could mentally “see” and “feel” the shapes described by the sounds.
The prototype has the potential to not only help the visually impaired, but to also benefit any user of a mobile device. In situations where a person could use the assistance of a mobile device, but cannot easily look at a screen; or in instances when not all visual information is present (such as in online gaming), the system can provide real-time guidance in the real world or a virtual world.
IBM researchers in Tokyo are advancing the prototype technology with the goal of helping empower the visually impaired to better-navigate the ever increasing non-textual, visual information on the web.