Archive for the ‘IBM’ Category
IBM’s Watson Hired For His First Real Job
IBM's Jeopardy! master robot Watson may not be anytime soon, but he has gotten his first job: as a diagnostic whiz, like . (Note: We will refer to Watson as a "he" and not an "it" until he stops being more charismatic than most humans we know.) According to the , IBM and health insurer WellPoint have agreed to use Watson to "help suggest treatment options and diagnoses to doctors." Congratulations, Watson! Don't blow your first paycheck on anything frivolous! []
Yale Law Journal Ponders the Wisdom of IBM Robot Watson as a Judge
The Honorable Justice Watson?
The Yale Law Journal's Betsy Cooper examining our favorite Jeopardy! champion (and ) robot Watson, but from a new angle: Could Watson help judges make legal decisions?
The essay notes that Watson could be of particular use to a certain type of judge or legal scholar: the new textualists. She writes: "New textualists believe in reducing the discretion of judges in analyzing statutes. Thus, they advocate for relatively formulaic and systematic interpretative rules. How better to limit the risk of normative judgments creeping into statutory interpretation than by allowing a computer to do the work?"
Says Cooper, "there are three important elements of new textualism: its reliance on ordinary meaning (the premise), its emphasis on context (the process), and its rejection of normative biases (the reasoning)." From that vantage point, Watson wouldn't be so much a judge (much as we'd love to see a massive black judge's robe draped over Watson's storage array) as an assistant or clerk, using its power to decide, for example, what the most "ordinary" use of a word is. Humans have to rely on instinct and experience, but Watson can systematically measure that sort of thing, narrowing down the possible meanings of words to eliminate uncertainty.
Watson also has the advantage of not being able to insert his own emotions or opinions into his decisions, by virtue of the fact that, well, he doesn't have any. Cooper does conclude that, due to his occasional errors (we'd hate to sentence criminals to serve time in Toronto) and the more basic fact that perhaps there should be a human element to judging, Watson is not an ideal candidate to actually make the bench. But that doesn't mean he couldn't be tremendously useful in legal decisions.
[]
IBM Is Building the Largest Data Storage Array Ever, 120 Petabytes Big
There are plenty of challenges inherent in building this kind of groundbreaking array, which, says, IBM, is destined to be used for, as writes, "an unnamed client that needs a new supercomputer for detailed simulations of real-world phenomena." For one thing, IBM had to rely on water-cooling units rather than traditional fans, as this many hard drives that can't be subdued in the normal manner. There's also a sophisticated backup system that senses the number of hard disk failures and adjusts the speed of rebuilding data accordingly--the more failures, the faster it rebuilds. According to IBM, that should allow it to operate with the absolute minimum of data loss, even none.
IBM's also using a new filesystem, designed in-house, that writes individual files to multiple disks so different parts of the file can be read and written to at the same time.
This kind of array is bottlenecked pretty severely by the speed of the drives themselves, so IBM has to rely on software improvements like that new recovery and filesystem to up the speed and enable the use of so many different drives at once.
Arrays like this could be used for all kinds of high-intensity work, especially data-heavy duties like weather and seismic monitoring (or )--though of course we're curious as to what this particular array will be used for.
[ via ]
New Computer Chip Modeled on a Living Brain Can Learn and Remember
IBM, with help from DARPA, has built two working prototypes of a "neurosynaptic chip." Based on the neurons and synapses of the brain, these first-generation cognitive computing cores could represent a major leap in power, speed and efficiency
About a year ago, we told you about IBM’s project to map the , the most complex brain networking project of its kind. Big Blue wasn’t doing it just for the sake of science — the goal was to reverse-engineer neural networks, helping pave the way to cognitive computer systems that can think as efficiently as the brain. Now they’ve made just such a system — two, actually — and they’re calling them neurosynaptic chips.
Built on 45 nanometer silicon/metal oxide semiconductor platform, both chips have 256 neurons. One chip has 262,144 programmable synapses and the other contains 65,536 learning synapses — which can remember and learn from their own actions. IBM researchers have used the compute cores for experiments in navigation, machine vision, pattern recognition, associative memory and classification, the company says. It’s a step toward redefining computers as adaptable, holistic learning systems, rather than yes-or-no calculators.
“This new architecture represents a critical shift away form today’s traditional von Neumann computers, to extremely power-efficient architecture,” Dharmendra Modha, project leader for IBM Research, said in an interview. “It integrates memory with processors, and it is fundamentally massively parallel and distributed as well as event-driven, so it begins to rival the brain’s function, power and space.”
You can read up on Von Neumann architecture , but essentially it is a system with two data portals, which are shared by the input instructions and output data. This creates a bottleneck that will fundamentally limit the speed of memory transfer. IBM’s system eliminates that bottleneck by putting the circuits for data computation and storage together, allowing the system to compute information from multiple sources at the same time with greater efficiency. Also like the brain, the chips have synaptic plasticity, meaning certain regions can be reconfigured to perform tasks to which they were not initially assigned.
IBM’s long-term goal is to build a chip system with 10 billion neurons and 100 trillion synapses that consumes just one kilowatt-hour of electricity and fits inside a shoebox, Modha said.
The project is funded by DARPA’s SyNAPSE initiative, and IBM just completed phases 0 and 1. IBM’s project, which involves collaborators from Columbia University, Cornell University, the University of California-Merced and the University of Wisconsin-Madison, just received another $21 million in funding for phase 2, the company said.
Computer scientists have been working for some time on systems that can emulate the brain’s massively parallel, low-power computing prowess, and they’ve made several breakthroughs. Last year, computer engineer Steve Furber described a that consists of tens of thousands of cellphone chips.
The most notable computer-brain achievements have been in the field of . As their name implies, a memory resistor can “remember” the last resistance that it possessed when current was flowing through it — so after current is turned back on, the resistance of the circuit will be the same. We will not attempt to delve too deeply here, but this basically makes a system much more efficient.
Hewlett-Packard has been since first describing them in 2008, and has also been part of the SyNAPSE project. Last spring, HP engineers described a titanium dioxide memristor that uses low power.
For a brain-based computer system, memristors can function as a computer analogue for a synapse, which also stores information about previous data transfer. IBM's chip doesn't use a memristor architecture, but it does integrate memory with computation power — and it uses computer neurons and axons to do it. The building blocks are simple, but the architecture is unique, said Rajit Manohar, associate dean for research and graduate studies in the engineering school at Cornell.
"When a neuron changes its state, the state it is modifying is its own state, not the state of something else. So you can physically co-locate the circuit to do the computation, and the circuit to store the state. They can be very close to each other, so that cooperation becomes very efficient," he said.
Modha said it is just a new way to store memory.
"A bit is a bit is a bit. You could store a bit in a memristor, or a phase-change memory, or a nano-electromechanical switch, or SRAM, or any form of memory that you please. But by itself, that does not a complete architecture make," Modha said. "It has no computational capability."
But this new chip does have that power, he said. It integrates memory with processor capability on a typical SOI-CMOS platform, using traditional transistors in a new design. Along with integrated memory to stand in for synapses, the neurosynaptic “core” uses typical transistors for input-output capability, i.e. neurons.
This new architecture will not replace traditional computers, however. “Both will be with us for a long time to come, and continue to serve humanity,” Modha predicted.
The idea is that future powerful chips based on this brain-network design will be able to ingest and compute information from multiple inputs and make sense of it all — just like the brain does.
A cognitive computer monitoring the oceans could record and compute variables like temperature, wave height and acoustics, and decide whether to issue tsunami or hurricane warnings. Or a grocer stocking shelves could use a special glove that monitors scent, texture and sight to flag contaminated produce, Modha said. Modern computers can’t handle that level of detail from so many inputs, he said. But our brains do it all the time — grab a rotting peach, and your senses of touch, smell and sight work in concert instantaneously to determine that the fruit is bad.
To do this, the brain uses electrical signals between some 150 trillion synapses, all while sipping energy — our brains need about 20 watts to function. Understanding how this works is key to building brain-based computers, which is why IBM has been working with neuroscientists to study monkey and . That research is progressing, Modha said.
But it will be quite some time before computer chips can truly match the ultra-efficient computational powerhouses that nature gave us.
IBM Mysteriously Halts Work on the World’s Fastest Academic Supercomputer
Usually we’d spend the second paragraph telling you why, but in this case we just don’t know. IBM said the supercomputer became more expensive and more complex than the company foresaw. A company spokeswoman said IBM is capable of meeting the technological goals outlined for the project, but nonetheless it is choosing not to.
That’s all a bit odd. The computer, known as Blue Waters, is a building-sized behemoth costing roughly half a billion dollars, much of which was funded by the National Science Foundation. It was based on IBM’s Power7 series chip that is not yet on the market. Which makes one wonder if there was a problem with the chip or with the architecture of the computer itself. Or maybe upon building the first few racks of hardware the computer started to think for itself (with few answers to work with, we’re taking license to speculate here).
But the world’s biggest, baddest academic computer isn’t necessarily lost. The National Center for Supercomputing Applications (you may remember it from our coverage of U. of Illinois’ earlier this year), which is heading the effort, is seeking other means to finish the computer without IBM. But it only has a few weeks to get another plan in front of the NSF.
Unfortunately, this kind of hardware doesn’t exactly exist in plug-and-play format, so we’ll have to wait and see if some other chip developer can step in and make the NCSA’s new supercomputer as super as it was supposed to be.
[]
IBM Data Analysis Platform to Plan a More Efficient Future, Coming to a City Near You
The Intelligent Operations Center is designed to aggregate data from multiple government IT systems, which could help city planners spot trends and connections among various types of data. As we reported previously, IBM began rolling out the porject in Rio de Janeiro, where algorithms are helping and coordinate emergency responses. IBM has also been working with smaller locales like Dubuque, Iowa, to study anything from public safety to water use.
The platform could integrate public transportation information with traffic management, for instance, letting cities plan better bus routes or traffic signal patterns based on congestion at various times of day. Or it could monitor maintenance logs on city infrastructure, deploying utility crews to fix water pipes or other assets before they break. It could even help planners devise better emergency responses, analyzing cameras and crime databases to catch criminals or even prevent crime, IBM says.
It will all be cloud-based, so cities won’t have to hire teams of on-the-ground IT consultants to sift through competing types of data. This will be cheaper for cities, but also easier for IBM, which can easily customize code for any municipality.
The plug-and-play system will have several modules based on various themes, like public safety or water management. None of the modules are available yet, but IBM will roll them out over the next year, explains. IBM engineers noticed several common threads in its previous Smarter Cities projects, like how to deal with congestion, and decided to turn them into code.
IBM plans to offer the software starting June 17. Cities will be able to contract with Big Blue itself or various vendors, the company says. No word yet on price, but IBM says the system's ability to spot inefficiencies will save cities money in the long run. For instance, Alameda County, Calif., used a software system to coordinate its social services, which saved $25 million a year, IBM tells PC World.
This could be a promising proposition for cash-strapped municipalities still reeling from the recession. And that sounds pretty smart.
[via , ]
In Brazil, an Explosion in Computing Power is Revolutionizing Weather Prediction
How better weather forecasts predict a more efficient future
Even a decade ago, something like this wouldn’t have been possible. But now, explosions in computing power and sophisticated software design are driving a revolution in atmospheric modeling that allows researchers to predict the weather and its impacts in whole new ways--ways that might not just save us from the occasional wrath of a violent storm, but also from ourselves.
“Since 1950 we’ve basically witnessed a revolution in weather forecasting,” says Dr. Louis W. Uccellini, director of the National Centers for Environmental Prediction at the U.S. National Weather Service, noting the proliferation of satellites in orbit and sensor networks on the ground and in the sky that now provide terabytes of global weather data. What’s been missing is a way to find meaning in all of this data. “Right now, the biggest weakness is the computational capacity that we’re able to bring to this enterprise,” he says.
That’s changing as supercomputing power becomes commonplace. The NWS itself is in the process of procuring its next-generation computing platform. More computing power means better weather models running on better algorithms crunching more data.
Uccellini offers a simplified example: whereas earlier generations of the NWS’s weather models relied generally upon various atmospheric data to make their projections, the NWS now commonly runs atmospheric models alongside land models, ice models, and ocean models. So while those earlier projections might predict that it was going to snow, Uccellini says, scientists can now mash up the atmospheric and land model data to predict if that snow is going to stick. The NWS’s next-gen computer should churn out even richer forecasts, with more layers of meaning.
If predicting the weather was our only concern, we could end the story there. But that’s rarely the case.
PREDICTING THE IMPACTS
“If all you can do is solve the weather problem, that’s insufficient,” says Lloyd Treinish, who runs the Deep Thunder initiative from IBM’s Yorktown, N.Y., research campus. “That doesn’t solve the business problem, which is connected to the impact of the weather.”
Deep Thunder isn’t new--the project has been modeling weather in the New York area for a decade--but it is pushing the current state of the art in weather forecasting. It is a computing solution designed to provide extremely accurate forecasts tailored to the specific needs and problems. It doesn’t provide particularly long-term weather projections--the further out a model attempts to forecast, the less solid data is available; even with all the computing power scientists could dream of, there will always be atmospheric variables that degrade the accuracy of forecasts the further out they are made--but out to about 48 hours it is remarkably precise.
What makes Deep Thunder and models like it so interesting--and potentially game-changing--is what Treinish calls coupling--the use of weather models to predict not just the weather, but the impacts of weather. For that, Deep Thunder uses data from NOAA, NASA, the NWS, the U.S. Geological Survey--as well as, uniquely, detailed data that’s not about weather per se, but that’s relevant to whatever particular problem the project is tackling.
Take the example of a power utility. Plug information relevant to that utility into Deep Thunder--where its power stations are, where each transformer is, how its service trucks are distributed throughout the area, historical grid data--and run it through Deep Thunder. The algorithms can tell you whether tomorrow’s incoming thunderstorm is likely to cause a power outage, and if so where. The utility can then reorient its resources to deal with the threat ahead of time, restoring power faster and keeping life (and commerce) humming along at a normal pace.
Extrapolate that scenario to any number of human weather-related problems, and it’s easy to see how these new models are going to reshape economies, cities, government agencies, and just about everything else that happens under the sky. With sophisticated weather models coupled to their particular problems, freight companies can minimize delays, governments could better plan for severe storms (like the snowstorms that crippled New York last winter), utilities could better plan for tomorrow’s energy loads, and agencies could even brace for specific disasters that at best interrupt life and at worst end it.
That’s what Rio de Janeiro's is all about. Working with IBM as part of its Smarter Cities initiative, local and regional authorities there are building a central information hub that ties the various stakeholders in Rio together: agencies, state and local governments, police, firefighters, sanitation, city services. All share information so they can quickly react to problems and execute their offices efficiently. Deep Thunder was one of the first technologies implemented there.
“There’s a very specific problem that they’re trying to address,” Treinish says. “These large storms that impact the city can easily dump a foot of rain in 24 hours, and when they’ve had events like this it leads to significant loss of life and significant property loss.”
In Rio, a one-size-fits-all approach to its unique topography and weather is of little use. The urban landscape there is unique, with huge shanty neighborhoods built into the steep hillsides surrounding the city proper.
“These storms come off the ocean and they interact with the very complex terrain in Rio, and potentially the water begins to accumulate on these steep hillsides and it leads to mudslides,” Treinish says. “So the large amount of rain and the flooding is a problem, but ultimately it’s the mudslides that lead to the societal and economic impacts in the city.”
In the Rio datacenter, Deep Thunder takes into account city-specific soil composition data, hydrology models, urban flooding models, topographical data, population density and land use data. With the resulting output, city authorities can pinpoint which hillsides are at the greatest risk of mudslides during a certain storm and allocate resources (and warn inhabitants) accordingly.
Such high-powered, coupled models won’t just solve problems for governments and businesses--they could potentially solve some of humanity’s greatest big-picture challenges as well.
Homes and buildings that can plan for the day ahead computationally will be astronomically more efficient than today’s “smartest” structures. For a “smart grid” to be smart, it first has to know what’s happening, and knowing--down to the mile, down to the building, down to the very person--what kind of energy is likely going to be drained from the grid in a given hour is key to making it work.
“Weather forecasts could drive the cooling and heating of your home, hands-off,” says John Bosse, director of energy and government services for Earth Networks, the company that runs WeatherBug (a maker of apps and desktop widgets that keep users updated on the weather) and overseer of one of the nation’s larger commercial weather sensor networks. “Most people just set their thermostat at a temperature. You could be so much more efficient. It’s not like you have to look at the forecast every morning and program your thermostat, this would all be automated.”
More precise weather modeling is also critical to enabling the rollout of renewable, carbon-free energy sources like wind and solar. Their resources are the weather, after all, but their inconsistency is the main obstacle keeping utilities from relying on them for more than an auxiliary role. If you know with a high degree of certainty how much sunlight and how much wind you’ll have from day to day, you can lean more heavily on those carbon-free options--and power down the coal plants when the weather is right.
This is a future that is going to take some time (and a lot of investment) to realize, but the seeds are already taking root in places like Rio. As weather sensor networks are integrated--NOAA has already begun stitching together America’s patchwork sensor networks, creating a network of networks that algorithms can pull from--and computing power continues to fall in price and proliferate, this kind of modeling will become commonplace.
It will become more accurate too, Treinish says, because those sensors will inform the algorithms and vice versa. “It’s the enabling of a feedback loop between the observations and the modeling and the simulations,” Treinish says. “Better weather observations enable better models, and better models can be used to determine additional ways to sample the atmosphere.”
How will all this impact your local morning weather report? “It’s probably not going to be striking,” Bosse says. “It’s not going to be night and day. What you are going to see is more precision and better accuracy in short term forecasts.” Meaning your local Roker will still stand in front of those radar maps and talk about pressure systems moving this way and that.
But you may notice that the hour-by-hour forecast predicting what time an afternoon shower will roll in becoming increasingly more accurate. That’s the algorithms at work, chipping away at atmospheric uncertainties and helping to build a smarter, more efficient future.