Technology Trends

I’m sure that many of you are sick and tired to carry their laptops during your trips. Of course, you can use a USB key ring to carry your data to access it from any other PC. But what about having a handheld device, such as an iPod, containing the full image of your computing environment, and restoring this whole environment on another PC anywhere in the world? Thanks to researchers from IBM, it is now possible to use the SoulPad system, a portable device carrying a stack of software. The host PC, which can be of any variety, “boots an auto-configuring operating system (Knoppix) from the SoulPad, starts a virtual machine monitor, and resumes a suspended virtual machine that has the user’s entire personal computing environment.” Now, the IBM team is thinking to use cell phones as well to carry the SoulPad system as soon as they have enough disk storage capacity. This day, it will feel easier to travel…

Before going further, here is a diagram showing the multi-tiered architecture of the SoulPad (Credit: IBM Research).

Here are more details from IBM Research.

Essentially, SoulPad enables a user to hibernate a PC session to a pocket form-factor device and carry the device to some another PC and resume his session on that PC. SoulPad has minimal dependencies on PCs that can be used to resume a user session.

In specific, PCs are neither required to be network connected, nor have any pre-installed software. The only requirement is the support of a high speed local connection to a SoulPad device for an acceptable suspend/resume times and acceptable runtime performance.

The researchers note that their approach is very different from the Intel’s Internet Suspend/Resume (ISR) project because SoulPad doesn’t require a known stack of software on the second PC.

As you might guess, the second PC doesn’t boot instantaneously. But suspending and restoring times are very similar to the time it takes for your laptop to move to hibernating mode, about 2 minutes using a USB 2.0 connection.

And what about security if your SoulPad is stolen?

To protect user data if a SoulPad is misplaced or stolen, we encrypt the disk partition that holds the VM images using the AES128 block cipher. We used the publicly available loop-aes package for Linux in our implementation.

The encryption key is generated by hashing a usersupplied passphrase. After the Host OS boots, it prompts the user to enter the passphrase. If the user supplies an incorrect passphrase, the resulting hash will not correspond to the AES key and the mount operation will fail since the decrypted data will not correspond to a valid filesystem. In order to defeat brute force attacks that attempt to guess the passphrase, the loop-aes package requires the passphrase to be at least 20 characters long. For convenience, we permit users to supply this passphrase via an auxiliary USB flash key.

But will a friend allow you to use his PC? You can tell him that the SoulPad system doesn’t touch anything on his machine and will not leave any traces either.

The SoulPad project was presented at MobiSys 2005, the Usenix Third Annual International Conference on Mobile Systems, Applications, and Services, which was held on June 6-8, 2005, in Seattle, WA.

The researchers received an award for Best Paper for “Reincarnating PCs with Portable SoulPads.” Here are two links to the abstract and to the full paper (PDF format, 14 pages, 187 KB).

The IBM Research site also provides a link to a video showing how the SoulPad concept works (5 minutes and 53 seconds, 14 MB).

Finally, you might want to read two other articles about the SoulPad project, “System carries PC soul” from Technology Research News and “Pocket-sized computer ’soul’ developed” from New Scientist.

Sources: IBM Research Project page about SoulPad, 2005; and various web sites

Related stories can be found in the following categories.

• Computers
  


• Handhelds
  


• IBM
  


• Innovation
  


• Pervasive Computing
  


• Software
  


• Technology
  

And remember that comments are no longer accepted here because of a vandal. If you want to tell me something about this post, please go to the bottom right of this page and send me an e-mail.

Computer scientists from Carnegie Mellon University (CMU) are developing and using online games to train computers to better see according to the Pittsburgh Post-Gazette. One of these games is Peekaboom, used online for free by teams of two players. The first one, designated as “Peek,” sees on his screen an image — initially empty — and a word that describes the image or one element of the image. The second one, named “Boom,” gradually reveals the image or gives hints to “Peek” until he correctly guesses the word associated to the image. And this use of “segmented” images might improve “computer vision” by “teaching” them how to identify objects. But read more before playing…

Below is a simplified trial run of Peekaboom, where “Boom” is the Revealer and “Peek” the guesser (Credit: Peekaboom.org).

You’ll find a larger version of this image — along with other details — on page 15 of this presentation named “Peekaboom: A Game for Locating Objects in Images” (Powerpoint format, 23 pages, 809 KB).

The images and words used in Peekaboom, created by Luis von Ahn, have been taken from a previous CMU online game, the ESP Game, in which participants generate descriptive labels for images.

Now, let’s read some short excerpts from the Pittsburgh Post-Gazette article, starting with the game itself.

In the game, which graduate student Roy Liu has been programming since September, one player is given an image and a word that describes the image or one element of the image. The object of the game is for the first player, designated as Boom, to get the other player, designated as Peek, to correctly guess the word.

Peek initially sees only a blank screen, but as play proceeds Boom reveals the image, little by little, by moving his cursor over the image.

When the word is correctly guessed, or when the players agree to pass on an image, the players switch roles and play resumes. The players receive scores based on the number of correct guesses they can make in four minutes.

But how can this game improve computer vision?

In the early days of machine vision research, it was assumed that computers could learn to identify an object, such as a car, or a spoon, or a face, if it was given the rough geometry of the object, said Alexei Efros, a computer vision researcher at Carnegie Mellon. But that wasn’t a successful approach.

A better approach to teaching a computer how to identify an object, such as a car, is to show it lots of images of cars, of various makes and colors, taken from a variety of angles and distances and under a variety of lighting conditions, he said.

For more information, you can check how Peekaboom works by reading this introduction (PDF format, 1 page, 461 KB).

But for fun, just play with Peekaboom!!!

Sources: Byron Spice, Pittsburgh Post-Gazette, August 1, 2005; and various web sites

Related stories can be found in the following categories.

• AI
  
  
• Computers
  
  
• Games
  
  
• Software
  
  
• Vision and Visualization Apps
  

And remember that comments are no longer accepted here. If you want to tell me something about this post, please go to the bottom right of this page and send me an e-mail.

Printed maps are easy to manipulate, provide an easy way of interacting for multiple users, but are static and can be out of date. On the contrary, computer-based map displays can provide dynamic and more recent information than paper-based maps, but do not help a group of people to communicate. So why not mix them? This is what have done researchers at England’s University of Cambridge with their augmented maps, which add digital graphical information and user interface components to printed maps. Here is how this works: the printed maps are placed on a flat surface; an overhead camera linked to a PC tracks the map via the live video stream; and an overhead projector adds graphical information to the maps. This could be useful for many applications, and the researchers have applied it to a flood simulation of the Cambridge area. Read more…

First, here is a diagram showing the whole system and its components (Credit: University of Cambridge, UK).

And below is an augmented map showing the flooded River Cam. “The image browser to the right shows views corresponding to locations and different stages of the flood, while the PDA to the left controls a helicopter unit” (Credit: University of Cambridge, UK).

Here is a description of the system which has been developed by Dr T.W. Drummond, Dr G. Reitmayr and Ethan Eade.

Tom’s demonstration of the dynamic paper map comprises of a camera and a projector looking down at a paper map from above. The system performs interactive tracking of the map on a table top environment using the live video stream captured by the camera. Once the locations of the maps are known, the projector displays extra information directly on the maps.

The system also tracks user interface devices which can be placed on the map and which enable access to information that is linked to locations on the map. A simple physical prop, for example a piece of white card, becomes a selection tool and projection surface at the same time. Images referenced by the location pointed at are displayed in the white card.

So far, Tom and his colleagues have used their system to show how it could be used to monitor a flooding situation in the Cambridge area and how easy it would be to deploy emergency units, such as an helicopter, by controlling it with a PDA.

Now, the researchers want to move out from their labs and build a deployable and mobile system.

You’ll find more information on the project page, with more technical explanations and different images.

For your viewing pleasure, here is a link to a short video (2 minutes and 43 seconds, 25.2 MB) showing the different tools and components of the system.

And if you’re interested by these augmented maps, a technical paper will be published soon under the name “Localisation and Interaction for Augmented Maps.” This paper will appear in the Proceedings of the 4th IEEE and ACM International Symposium on Mixed and Augmented Reality (ISMAR 2005), which will be held in Vienna, Austria, on October 5-8, 2005.

Sources: University of Cambridge, Engineering Department, News & Features, July 7, 2005; and various web sites

Related stories can be found in the following categories.

• Computers
  
  
• Engineering
  
  
• Human Computer Interface
  
  
• Innovation
  
  
• Vision and Visualization Apps
  

Admit it, typing an SMS on a cell phone takes time, and writing an e-mail on a PDA is only marginally better. But according to the San Jose Mercury News, a researcher at IBM has found a solution to this vexing problem. Instead of typing words on these ridiculous small keyboards, with the SHARK, an abbreviation for ShortHand-Aided Rapid Keyboarding, you use a grid and a stylus. The grid appears on the screen of your portable device. You put a stylus on the first letter of the word you want to type. Then you drag the stylus to draw a line connecting all the other letters of the word. When you release the stylus, the word appears almost magically. With SHARK, you can type between 50 and 80 words per minute, which is almost miraculous. So far, IBM hasn’t yet decided to release this software as a product. But if enough of you download it, which is currently free, and say you want it, IBM could release it as a paying product within a few months.

Here is the introduction of the Mercury News article.

Humans in their long history have invented only two ways for individuals to produce text: handwriting and typing on a keyboard.

Shumin Zhai, an IBM scientist, may have invented another way: SHARK, an abbreviation for ShortHand-Aided Rapid Keyboarding.

SHARK is intended for writing text with a stylus on small touch-sensitive screens, such as those found in cell phones and personal digital assistants. It uses a radically different approach that is easy to learn and fast.

Here is how the system works. Below is a screen capture of a user trying to finishing to type “The quick brown fox jumps over the lazy dog.” On this capture, the user is moving its stylus to create the word “jumps” (Credit: IBM Almaden Research Center).

If you want to see SHARK in action without downloading it, here is a link to a video demo (4 minutes and 24 seconds, 29.7 MB). The above image comes from this video.

Here is how the Mercury News describe the system.

To write a word, you put the stylus on the first letter of the word and then drag the stylus to draw a line through the alphabet cluster, touching every letter in the word. When you lift up the stylus after hitting the last letter, SHARK figures out what word you want and displays it on the screen.

If SHARK makes a mistake, you tap the word and get a list of the most likely alternatives based on the path you traced through the grid.

You can check the system by yourself, and even download a beta version on the IBM SHARK Shorthand web site.

CNET News.com also described the SHARK system last week in “New-age keyboard: Trace, don’t write.”

But for more technical information, here is a link to the recent publications of Shumin Zhai and his colleagues.

In particular, you should read “In Search of Effective Text Input Interfaces for Off the Desktop Computing” (PDF format, 18 pages, 255 KB).

For the moment, the system is only working with a database of English words. If IBM ever needs beta testers for a French version, I’m available. Typing text messages is just a nightmare right now…

Sources: Mike Langberg, San Jose Mercury News, July 15, 2005; and various web sites

Related stories can be found in the following categories.

• Computers
  


• Human Computer Interface
  


• IBM
  


• Innovation
  


• Software
  


• Technology
  
  
• Wireless
  

Traditional Oriental ink painting is more easily done with real brushes than with a computer program because you need to model how the ink is flowing into an absorbent surface such as paper. In this brief article, Technology Research News writes that “researchers from the Hong Kong University of Science and Technology have developed a brush-and-ink-style paint program, dubbed MoXi, that uses a model of pigment particles in water flowing into paper.” These virtual Chinese brushes simulate in real time the ink dispersion and could be available on your PC within two years. Read more…

Here is some general information about MoXi provided by Technology Research News.

The software models the gritty details of paper absorbing water and pigment moving through water, including the way pigment concentrates at ink boundaries as water evaporates from drying ink. The technique promises to make computer paint programs with more realistic and could also be used in computer animation packages, according to the researchers.

The simulation is based on mathematics — the lattice Boltzmann equation — that physicists use to model the complex behaviors of fluids. The model simulates more complex effects than previous work, and is also fast enough to deliver ink dispersion simulations in real-time on a reasonably large canvas, according to the researchers.

Below are two images generated with MoXi, the first one being called “Lotus leaves” and the second one “Planet” (Credit: Hong Kong University of Science & Technology)

Here are two links to larger versions of these images, the “Lotus leaves” (1.30 MB) and the “Planet” (1.47 MB).

The researchers behind the MoXi project are Chiew-Lan Tai, Associate Professor at the Department of Computer Science, and Nelson Siu-Hang Chu, her Research Assistant.

For more information about their projects, you can read these two pages about the Virtual Chinese Brush and about MoXi. On this page, you’ll have access to several videos and images. The two pictures above come from this page.

The MoXi project will be presented at SIGGRAPH 2005 under the name “MoXi: Real-Time Ink Dispersion in Absorbent Paper.” Here is a link to the paper submitted by the researchers (PDF format, 1 page, 145 KB). Here are an excerpt from the introduction.

Our paint system, MoXi, allows users to paint in the spontaneous style of Eastern ink painting, on a computer. The simulations of both brush and ink are essential for a successful extension of this traditional art into the digital domain. For real-time performance, we have implemented our ink flow model entirely on the GPU, leaving the CPU for the brush simulation.

According to the researchers, this technique “could be used practically in one or two years.” But is this possible that this technology can be sold under the name MoXi? There already is a Digeo service named Moxi which offers High Definition TV (HDTV). And Digeo claims in its press releases (check this one for example) that Moxi is one of its registered trademarks.

However, it’s not so clear. I visited the United States Patent and Trademark Office (USPTO) to know more. And for more information about this trademark, you can either click on the “Status” button in the Trademarks section, and enter the serial number 76279215 on the next screen, or go directly here. Here is the status of this trademark application as of April 25, 2005.

A non-final action has been mailed. This is a letter from the examining attorney requesting additional information and/or making an initial refusal. However, no final determination as to the registrability of the mark has been made.

If I correctly understand English, this means that Moxi IS NOT a registered trademark. But at the same time, Digeo writes it is registered. Who is right?

If one of the readers of this note is familiar with the USPTO procedures, please post an explanation below. Thanks.

Sources: Technology Research News, June 29/July 6, 2005; and various web sites

Related stories can be found in the following categories.

• Computers
  


• Graphics
  


• Mathematics
  


• Patents
  


• Software
  

We’re using computers for so long now that I guess that many of you think that our brains are working like clusters of computers. Like them, we can do several things ’simultaneously’ with our ‘processors.’ But each of these processors, in our brain or in a cluster of computers, is supposed to act sequentially. Not so fast! According to a new study from Cornell University, this is not true, and our mental processing is continuous. By tracking mouse movements of students working with their computers, the researchers found that our learning process was similar to other biological organisms: we’re not learning through a series of 0’s and 1’s. Instead, our brain is cascading through shades of grey. Read more…

According to this study, learning — at least language comprehension — is a continuous process.

“For decades, the cognitive and neural sciences have treated mental processes as though they involved passing discrete packets of information in a strictly feed-forward fashion from one cognitive module to the next or in a string of individuated binary symbols — like a digital computer,” said Michael Spivey, a psycholinguist and associate professor of psychology at Cornell.

His experiments are somewhat fascinating — even if limited.

In his study, 42 students listened to instructions to click on pictures of different objects on a computer screen. When the students heard a word, such as “candle,” and were presented with two pictures whose names did not sound alike, such as a candle and a jacket, the trajectories of their mouse movements were quite straight and directly to the candle.

The picture below shows Michael Spivey with one of his students looking at two objects on her screen.

[He asked her] to listen for a word and then to click on its picture. By studying the curvature of the trajectory of the mouse, he can analyze language comprehension processes (Credit: Kevin Stearns, Cornell University).

But when the students heard “candle” and were presented with two pictures with similarly sounding names, such as candle and candy, they were slower to click on the correct object, and their mouse trajectories were much more curved. Spivey said that the listeners started processing what they heard even before the entire word was spoken.

Spivey concludes that our brains can handle ambiguities.

“When there was ambiguity, the participants briefly didn’t know which picture was correct and so for several dozen milliseconds, they were in multiple states at once. They didn’t move all the way to one picture and then correct their movement if they realized they were wrong, but instead they traveled through an intermediate gray area,” explained Spivey.

For more information, the research work has been published online by the Proceedings of the National Academy of Sciences under the name “Continuous attraction toward phonological competitors.” Here is a link to the abstract.

Certain models of spoken-language processing, like those for many other perceptual and cognitive processes, posit continuous uptake of sensory input and dynamic competition between simultaneously active representations. Here, we provide compelling evidence for this continuity assumption by using a continuous response, hand movements, to track the temporal dynamics of lexical activations during real-time spoken-word recognition in a visual context. By recording the streaming x, y coordinates of continuous goal-directed hand movement in a spoken-language task, online accrual of acoustic-phonetic input and competition between partially active lexical representations are revealed in the shape of the movement trajectories. This hand-movement paradigm allows one to project the internal processing of spoken-word recognition onto a two-dimensional layout of continuous motor output, providing a concrete visualization of the attractor dynamics involved in language processing.

The access to the full article will cost you $10.

Now, I have a question for you. Even if this new study is right, what will it change for us? Will you wake up differently tomorrow morning? I don’t think so.

Sources: Susan S. Lang, Cornell News Service, June 27, 2005; and various web sites

Related stories can be found in the following categories.

• Computers
  
  
• Medicine
  
  
• Psychology
  
  
• Science
  

Like many of you, I had a good laugh a month ago when I read that some students at the MIT submitted a computer-generated ’scientific’ paper to a computer conference which accepted it, at least in a first step. (See ‘Prank research paper makes the grade‘ for example.) But now, I’m not laughing anymore. Imagine that 100,000 people around the world use this Automatic CS Paper Generator to generate a fake paper and keep it online. In our world of ‘permanent’ information, what will happen in five years when someone uses a search engine looking for keywords contained in the title of these fake papers? One of these papers may appear high in the list of results and this person may use this computer-generated paper as a basis for one of his projects. Scary, isn’t? Read more…

Let’s first go back to the original story in case you missed it.

On April 15, 2005, the MIT News office wrote that some MIT computer science students were so tired to see their papers rejected by scientific conference people that they started to have some doubts about their standards to accept or refuse a paper. (see link above for more details.)

So they decided to have some fun and to write software that generates meaningless research papers and submit them to different organizations.

One of their computer-generated papers, “Rooter: A Methodology for the Typical Unification of Access Points and Redundancy” (PDF format, 4 pages, 92 KB) was initially accepted by World Multi-Conference on Systemics, Cybernetics and Informatics 2005 (WMSCI 2005) as a non-reviewed paper, and later rejected.

Now, let’s have some fun and build a meaningless computer science paper. It’s very easy. On the site mentioned above, you just fill the names of one to five authors and submit your request. That’s all for you to do!

As an example, here are two fake papers that I ‘co-authored’ with some well-known people in the computer industry.

• “Boolean Logic Considered Harmful” by Linus Torvalds, Bill Gates and Roland Piquepaille (PDF format, 6 pages, 95 KB)


• “A Study of XML Using AcridLamb” by Paul Otellini, Roland Piquepaille and Hector Ruiz (PDF format, 3 pages, 46 KB)

It’s pretty easy to imagine a group of people, with fun or evil intentions, to link to such a computer-generated document in order to see it ranked high by search engines. If enough people are putting a link to the first document mentioned above, a Google search for ‘boolean’ and ‘harmful’ will soon return this fake document as its #1 result.

Of course, I don’t see why people would do that. And the probability that a real paper was co-authored by Bill Gates and Linus Torvalds is very low, so I don’t think anyone will think it’s a genuine document.

But lots of ‘phishing’ attacks these days show that people are more gullible than we might think.

So, is the possibility of hundred of thousands of fake computer science papers sitting online represents a danger or not? Time will tell, but please let me know what you think.

Sources: Roland Piquepaille, with various websites

Related stories can be found in the following categories.

• Computers
  


• Science
  


• Search
  


• Software
  

The U.S. Army is increasingly using battlefield sensors these days, from static ones which detect ground movements to unmanned aerial drones. This means that field commanders are overwhelmed by too much information. So the military started the BrUte Force Fusion (BUFF) program, which uses server and visualization systems from SGI, according to Military Information Technology. In a lab located at Fort Huachuca, AZ, BUFF analyses 170,000 hourly intelligence reports or about 3 terabytes of data each day. The goal of the BUFF program is to reach the Level II of the data fusion hierarchy — Level I occurs when a sensor is able to detect the movement of an object in a battle space while Level II blends data from multiple sensors. Many scientists believe that reaching Level II will take two decades, but promoters of the BUFF program think they can do better. Read more…

Here is the introduction of the article from Military Information Technology.

The Battle Command Battle Lab at Fort Huachuca, Arizona, is evaluating ways to fuse intelligence data from multiple sources into a cohesive combat picture in order to put more complete and actionable battlefield knowledge in the hands of field commanders. Experts there are using a variety of advanced technologies, including new server and visualization systems from Silicon Graphics Inc (SGI).

The new systems are deployed as part of the military’s Brute Force Fusion (BUFF) program, which is evaluating methods for achieving a clear understanding of current battlefield conditions. BUFF is focused on what is known as Level II fusion, which merges data from a growing array of intelligence sensors to create a picture of what is happening on the battlefield now.

With some 170,000 hourly intelligence reports generating three terabytes of data a day, BUFF’s research is also focused on assembling the data into actionable intelligence for specific field commanders or troop leaders.

In “Get BUFF,” GovExec.com gives some additional details about the program.

Jason Denno, the Battle Lab’s deputy director and a self-described “deviant” thinker, is leading the BrUte Force Fusion Program, a daring and potentially fruitless attempt to conquer the data deluge by wrestling it to the ground. (The muscular approach prompted the program’s acronym, “BUFF,” which was Denno’s idea.)

Denno hopes that, with BUFF, the military could produce level II fusion tools within the next few years. “We have a sea of information at this point,” Denno says. Buried in it, somewhere, could be the telltale clue that lets analysts know whether they’re looking at a few tanks moving across the desert, or the point of a much larger force, backed up by devastating artillery power.

Denno thinks that ‘traditional’ military analysts have not looked at enough information, probably because of a lack of computing power.

They have looked at the most current set of facts and assessed what was happening at that moment. When they have to make another assessment - probably a few minutes or a few days later — they look at the freshest data obtained since the last set.

The problem, Denno says, is analysts’ views frequently are based only on the most recent data; they might not account for historic trends. BUFF, however, would base each assessment on all the data that has been collected. Of course, that means the data set grows exponentially with each new assessment. The mound of data becomes a mountain and then a mountain range.

And here is Denno’s conclusion.

“Sometimes, you just need a sledgehammer,” Denno says of his all-or-nothing concept. “We were the first ones to use a sledgehammer instead of a pickle fork.”

For slightly different information, you also can read this SGI press release from February 15, 2005.

[Disclaimer: I worked for SGI between 1996 and 2001, but I don't have any more ties with this company.]

Sources: Harrison Donnelly, Military Information Technology, April 24, 2005; Shane Harris, GovExec.com, April 15, 2005; and various websites

Related stories can be found in the following categories.

• Computers
  


• IT
  


• MilitaryApplications
  


• Sensors
  

It took almost thirty years to get decent speech recognition programs on our computers. But if they’re good enough to translate our words into characters, they can’t engage in a conversation with us (I must say that some humans can’t do either). But according to this article from Technology Research News, things are changing. Computer scientists from Scotland and California have designed a multithreaded system which can anticipate what you’re going to say and are also able to switch context when you jump from a topic to another. This approach, which could be used in a wide range of applications, is welcome. Unfortunately, these researchers have selected the name “Conversational Interface Architecture” for their system, which leads to the worrisome acronym CIA. Anyway, the first commercial applications should be available within two years. Read more…

Here is a general description of this dialogue management system.

Researchers from Edinburgh University in Scotland and Stanford University have built a dialogue management system that promises to improve verbal communication with computers by giving the machine a sense of the type of phrase a person is likely to say next.

The Conversational Interface Architecture goes beyond the slot-filling dialogue systems commonly used for airline ticket booking systems by tracking multiple conversation threads, said Oliver Lemon, a senior research fellow at Edinburgh University. Slot-filling dialogue systems prompt users to provide topic-specific information and listen for keywords that determine the system’s response to the user.

And here ere are some details on how this dialogue management works.

The software follows multithreaded conversations — those that switch back and forth between several topics — without having to be programmed, regulates particular topics, and uses this information to improve speech recognition rates, according to the researchers. It also recognizes corrective fragments — phrases that correct something a user has just said — and it allows users to initiate, extend and correct dialogue threads at any time.

The system accomplishes this by tracking different types of utterances, including yes or no answers; who, what, where answers; and corrections like “I meant the office” and “not the tree.”

[Note: An utterance is a complete unit of talk, bounded by silence.]

I’s interesting to note that, by using this analysis of utterances, the system can work with any speech recognition system.

What could we do with such a software?

The approach could be used in a wide variety of speech recognition systems including telephone-based information systems, interactive entertainment devices, robots, computer interfaces for the visually impaired, in-car dialogue applications, and speech interfaces for personal computers.

Another question remains: when will such systems be available?

The context-sensitive component of the researchers’ system could be applied to practical applications now, said Lemon. Multithreaded dialogue management could be used practically within two years, he said.

This research work has been presented at the ACM Transactions on Computer-Human Interaction (TOCHI) conference last year and published in its September 2004 issue (Volume 11, Issue 3, Pages 241 - 267).

Here is a link to the abstract of this paper named “Multithreaded context for robust conversational interfaces: Context-sensitive speech recognition and interpretation of corrective fragments.” Here is a summary of their results.

In an evaluation of a dialogue system built using this architecture we found that 87.9 percent of recognized utterances were recognized using a context-specific language model, resulting in an 11.5 percent reduction in the overall utterance recognition error rate, and a 13.4 percent reduction in concept error rate. Thus we show that by using context-sensitive recognition based on the predicted type of the user’s next dialogue move, a more flexible dialogue system can also exhibit an improvement in speech recognition performance.

Sources: Eric Smalley, Technology Research News, April 6/13, 2005; and various websites

Related stories can be found in the following categories.

• Computers
  


• Human Computer Interface
  


• Innovation
  


• Software
  

Some students are luckier than others — or have more fun. For example, this Stanford University report says that some of the students there may have some hard and physical work to do: dancing. But in exchange, they’re working with sensors, cameras and computers to study how a dancer of the Merce Cunningham Dance Company is moving. This must be exhilarating, especially after finding — and confirming — that he acts as a ‘biomechanical rebel.’

Here is the experience of Jonah Bokaer, a dancer from the Merce Cunningham Dance Company, who was enrolled in the program.

The test subject danced wearing only blue shorts and the 50 silver balls the size of marbles that stuck to his skin, mapping out his physique.

“I know what I think my body is doing. But is it really doing that? I don’t really know, but I’d like to,” he said during a break in the afternoon session at the Motion and Gait Analysis Laboratory at Lucile Packard Children’s Hospital.

A member of the Merce Cunningham modern dance company, Jonah Bokaer said he couldn’t wait to see the results — a digital record of his skeleton’s behavior as it undulates, spins and leaps.

His moves are monitored by students of the Anatomy of Movement class which is now in its second year.

“We’re looking upside down, inside out, at the human body,” said course director Amy Ladd, MD, professor of orthopedic surgery. “It’s not the way any single discipline would frame the study of movement.”

Ladd added, “Each project reflects an integration of disciplines spanning the humanities and sciences to portray human movement.” The exercise was part of an extensive series of interdisciplinary art projects that were tied to Cunningham’s performances on campus last week.

So what methods are using these students to analyze a dancer’s movements?

Eight cameras in the lab tracked the motion of the silvery balls on their test subjects: Cunningham dancers Frank and Bokaer and course director Ladd, who also happens to be a trained ballet dancer.

“We thought that the study needed a comparison, and analyzing someone in pointe shoes would be a good contrast,” said Ladd, who has studied ballet for years. “So I reluctantly agreed.”

The cameras sent the data to a computer, operated by motion analysis lab’s engineer Erin Butler. The output includes motion capture of dancers as well as quantitative information.

But what do you learn from such interdisciplinary projects?

Projects like this, mixing science with art, are challenging to conceptualize, said Ladd. “We’re looking for projects that merge science and art. No one really knows how to do this well yet. It’s a difficult mix. It calls for a philosophical paradigm shift for people who have been trained to think in one realm or the other.”

Here is a link to the other projects at the Anatomy of Movement.

And as a conclusion, it’s not the first time that Stanford University is mixing several disciplines, such as arts, sports and science. Check for example this article from Technology Research News, “Sensors track martial arts blows.”

Sources: Rosanne Spector, Stanford University Report, March 16, 2005; and various websites

Related stories can be found in the following categories.

• Arts
  
  
• Computers
  
  
• Education
  
  
• Human Computer Interface
  
  
• Sensors
  

A friend of mine who worked for free to help senior citizens to use computers once told me that the biggest hurdle was not technical — people can learn during all their lives — but physical. Many old people have trembling hands which prevent them to use a mouse to point and click on a small icon on a computer screen or a link on a browser page. Now, according to this article from ExtremeTech, IBM has unveiled a mouse adapter which treats these tremors as “noise” by filtering out the unintentional movements of the hand caused by a tremor. This new mouse will also help the ten million people which are affected by this genetic disorder every year, and who aren’t necessarily old. This adapter will be sold for about $100. Read more…

Let’s start with some pictures.

Here is how it works.

The new mouse treats the hand tremors as noise, and uses algorithms based on image-stabilization systems used in digital cameras.

[As you can see on the above pictures,] the mouse includes a physical dial to control the sensitivity of the mouse, as well as how quickly the user needs to double-click. Normally, these functions are handled by software controls — which require a mouse to adjust.

As I wrote above, this inability to precisely use a computer mouse doesn’t affect only the elderly.

Although tremors are usually associated with the elderly, a type of tremor called Essential Tremor is actually a genetic disorder that affects 10 million people per year, according to the International Essential Tremor Foundation (IETF).

This mouse adapter will be distributed by Montrose Secam, a British electronics company. You can buy it now for £67.50, 119.00 euros or $107.00, depending on where you live.

For more information, you can read these two articles from the Mercury News, “Algorithm box smoothes hand tremors on mouse,” and from the San Francisco Chronicle, “Helping hand for those with shaky hands.”

Finally, you might want to read this IBM press release, “Mouse adapter gives computer access to millions of hand tremor sufferers,” which offers additional details and links.

Sources: ExtremeTech Staff, March 14, 2005; Therese Poletti, Mercury News, March 14, 2005; Benjamin Pimentel, San Francisco Chronicle, March 14, 2005; and various websites

Related stories can be found in the following categories.

• Computers
  
  
• Human Computer Interface
  
  
• Innovation
  
  
• Medicine
  

Intel President Paul Otellini, who will become CEO in May, has a blog. But it’s an internal blog, available only to Intel employees. Of course, there are always information leaks in companies the size of Intel (85,000 employees). So it’s not a surprise that the contents of Otellini’s blog are now appearing. In “Intel blog: what the boss thinks,” the Mercury News looks at this blog and also at other executives who blog directly for the public. The Mercury News also posted on its site the contents of Otellini’s blog (PDF format, 16 pages, 121 KB) for the period going from December 14, 2004 to January 19, 2005. Read more for selected excerpts.

Here is the introduction of the Mercury News article.

Like many a corporate executive, Intel President Paul Otellini rarely deviates from the company line in public. But read Otellini’s blog and you’ll see what he really thinks.

In contrast to the public online journals of most corporate executives, Otellini’s blog is strictly for Intel employees’ eyes only, published on the company’s internal Web site. But a copy of the 8-week-old blog obtained by the Mercury News shows Otellini unplugged, conversing candidly with Intel’s rank and file about the challenges facing the Santa Clara chip giant. Otellini, who becomes Intel’s next chief executive in May, even praises Advanced Micro Devices, his Sunnyvale rival whose existence company executives take pains to ignore.

If you read the Otellini’s blog as published by the Mercury News, you’ll notice that there are only four entries from Intel’s President. It’s more a weekly memo than a daily blog. Of course, I can understand this. Being Intel’s President and next CEO certainly guarantees a long weekly working time.

In his second entry, Otellini mentions that he was overwhelmed by the response to his blog, with 350 comments posted in the first 24 hours after publication. The file provided by the Mercury News lists only 16 comments to the first entry. This means that comments are certainly edited, and probably because many comments carry identical messages.

Now, here is my personal selection of Otellini’s writings, starting with his first post where he recognizes that his internal blog will eventually become public.

Why am I doing this? Well, it seemed like a good idea to be able to create an ongoing vehicle to share my thoughts and observations on Intel and our industry with our employees, and to allow you an opportunity to have a platform for your thoughts or responses. While this is intended as an internal blog, I recognize that it will become public — welcome to the Internet! As a result, please recognize that I may be a bit limited in my comments and responses to protect Intel, and that we may exercise some editorial privilege on your comments for the same reason. I want to be clear on this up front. This is the price of entry to this blog.

On December 19, Otellini wrote some comments about the Itanium family of microprocessors.

Yes, Itanium is taking longer to achieve market and financial success than we first envisioned. But we are making significant headway in the replacement of generations of RISC microprocessors with Itanium. This year, Itanium will out ship every RISC processor except Power or Sparc. Roughly 20% of the world’s supercomputers are now built on Itanium. I always felt it would take at least a decade for us to establish this architecture. It simply takes that long to establish the performance roadmap, and to develop a critical mass of software and users. We will stay the course on this product line, and we will succeed.

In his January 10 entry, he spoke about the tsunami in South Asia, but also about competition from AMD.

We have a significant competitor to our Xeon product line in the Opteron line from AMD. This is the first significant competition we have seen with the market segment for “high volume servers” that Intel invented. We identified this competitive threat over a year ago, and did a great job in 2004 by shipping record Xeon volumes, achieving a very rapid ramp of our 64 bit extension based product, and minimizing share loss. While I hate losing share, the reality is that our competitor has a very strong product offering. We did a good job in 2004, and cannot let up, as I expect 2005 to be even more competitive in this area. There is no quick fix or easy answer here.

I doubt he ever said this to a customer.

In the latest note, dated January 19, Otellini comments the reorganization of Intel into five divisions and about how some employees read about it on the Web before being informed internally.

Even with this “beat the web” approach, some employees have mentioned that they read about in on the web before they heard about it from Intel. My apologies, but it seems that it is very hard to move faster than internet speed. In any event, we are off and running on creating what one analyst called Intel 3.0. He talked about us originally being the Memory company, then the Microprocessor company, and now the Platform company. While there is some accuracy in this statement, I think it may be a bit premature to declare this as a done.” We have much work to do before we can truly claim to have capitalized on this fundamental shift in our business. I found many of the analyst comments regarding the reorganization to be very interesting.

Just in case you forgot about Intel’s reorganization last January, please read this article from CNET News.com, “Intel launches broad reorganization.”

So, is this a real internal blog or just a weekly memo to employees? I don’t think this is important. What matters is that every Intel employee can comment on the company’s strategy.

Sources: Dean Takahashi, Mercury News, February 15, 2005; Paul Otellini’s blog; Steven Musil and Michael Kanellos, CNET News.com, January 17, 2005

Related stories can be found in the following categories.

• Blogs
  
  
• Computers
  
  
• Intel
  
  
• Management
  

Last week, I was reading this press release from a company claiming it has developed the first mathematical derivation of Amdahl’s Law. The Colorado-based company, Massively Parallel Technologies, Inc. (MPT), doesn’t offer a paper about this claim on its web site. Instead, it says its software “makes supercomputing faster and cheaper than ever before.” MPT adds that any cluster using its software can achieve almost 100% efficiency. For example, they say they operate a 255-processor cluster “that accelerates difficult problems involving heavy cross-communication by an astounding 250 to 350 times” or that a “cluster with 1023 nodes requires over 40,000 times less communication time steps than its conventional counterpart.” The company looks real, but its claim seem extravagant. Has anyone used this technology? Is there something real behind these claims? Read more and post your comments…

Let’s start with two paragraphs from the gibberish MPT press release

The Massively Parallel approach, now confirmed by Amdahl’s law as the most efficient method of parallel processing, concentrates on improving the efficiency of the communication methodology between processors. This has resulted in systems that exceed 90 percent efficiency rather than the previously believed limit of 20 percent. The breakthrough also allows systems to be built with low cost commodity hardware to solve problems in minutes rather than hours.

Impressed with MPT’s achievements, Dr. Amdahl now sits on the company’s board of advisors. “It is a great pleasure for me to see the fruits of my work being realized so dramatically by MPT,” said Amdahl. “MPT’s technology will allow supercomputing power to be used by anyone, anywhere and I have no doubt it will result in dramatic achievement and scientific advancements. The sky’s the limit.”

OK, let’s switch from the press release to what the company says about its technology.

One of the opening paragraph sounds reasonable.

There are three primary data movement times which are important to parallel processing. They are data input, data output, and cross-communication. HOWARD [,their trademarked software,] has several novel solutions for I/O. One solution balances the time it takes to “cascade” a problem called ‘ψ’ with the amount of time it takes to perform the I/O of that problem ‘# time-steps’. HOWARD does this via the use of multiple communication channels ‘ψ_base‘ and MPT’s proprietary “Howard Cascade.”

But later down through the page, things start to appear as unreal. MPT says for instance that “in order for a conventional system with 41,391 nodes relying upon fast connectivity to equal the performance of a HOWARD system using office LAN quality connectivity (100Mb/s), it would need channel speeds of 15 Tb/s.”

By the way, do you know many conventional systems with more than 40,000 nodes exist today? None — if we except the soon-to-be-finished IBM’s Blue Gene/L system.

It’s time to switch to the MPT’s applications page, where the company explains its secret.

The secret is in MPT’s revolutionary new approach to parallel processing. HOWARD, unlike conventional architectures, does not rely upon fast low latency communication channels to connect processors in order to achieve the parallelism necessary to dramatically boost processing performance. Instead HOWARD is based upon an entirely new mathematical geometry in the distribution of problems to multiple processors and for cross-communicating partial solutions from one processor to another. The geometry utilized in a HOWARD cluster of 1023 nodes, for example, can utilize office LAN quality 100 Mb/s Ethernet connections for its channels and yet outperform conventional clusters with expensive 10 GB/s connectivity. In fact, for the most difficult of all cross-communication exercises, an “all to all” exchange between processors, the geometry of a HOWARD” cluster with 1023 nodes requires over 40,000 times less communication time steps than its conventional counterpart.

And MPT gives an example about the oil and gas industry, a sector I worked with for a number of years.

HOWARD is already revolutionizing the oil and gas industry. MPT currently operates a 255-processor cluster that processes seismic data so rapidly that it can produce nearly five million dollars worth of finished seismic data per year at an annual operating cost of just $250,000.

And if these fabulous numbers haven’t convinced you yet, what about the costs of an MPT solution?

The price per processor in a conventional parallel processing system is usually in the $10,000 to $100,000 range, while the efficiency of the system is usually under 10%. Thus a solution on a conventional system costs 40 to 400 times the price that it would cost if delivered on a single processor workstation in which the total cost of the system is around $2,500.

How is it possible that such a company has been ignored by the media — if its claims are true?

By the way, they really filed an application for a trademark for HOWARD. You can check with the United States Patent and Trademark Office. MPT files its “computer software in the field of parallel processing” under number 78513638 on November 9, 2004. But a trademark doesn’t imply that’s a product is good.

I’ve been involved for twenty years in supercomputing, working with many different parallel systems. And I never saw something similar to MPT’s claims. Is the company on something real? I’ve serious doubts, even if MPT received some grants from DARPA. If you ever used their software, or heard about it, please post your comments.

Sources: Massively Parallel Technologies press release, January 18, 2005; and MPT website

Related stories can be found in the following categories.

• Computers
  
  
• IT
  
  
• Software
  
  
• Supercomputers
  

A few months back, in “Will Azul Reign on the Server Market?,” I wrote about Azul Systems and its new approach to computing named ‘network-attach processing,’ similar to the NAS approach for data storage. Now Shahin Khan, VP and CMO at the company, has written a rather provocative article for The Register, “Get ready to buy chips by the kilo.” He argues that we soon should be prepared to order CPUs by the thousands and be ready for some new language. “Do we say: 2.5 kilo CPUs? Do we call this kilo core, or mega core processing? And since it goes way past current multi-core technology, do we call it poly-core technology?” Jon Udell, from InfoWorld, also commented Khan’s views in “VM-enabled polycore computing.” Read more…

Basically, Shahin Khan says that we are now used to very large numbers when we talk about memory or disk space for servers, but that the vast majority of us is still counting processors using single digits.

But when it comes to CPUs, we still mostly dabble in single digits. An 8-way server feels like a pretty large system. The 32-way, 64-way, and 200-way systems feel just huge. Even when we scale out, anything beyond a couple of hundred CPUs begins to challenge our ability to manage and operate the systems. It’s no accident that they call these systems a “complex.”

A major shift is coming. Over the next few years, your ordinary applications will be able to tap into systems with, say, 7,000 CPUs, 50 tera bytes of memory, and 20 peta bytes of storage. In 2005, Azul Systems will ship compute pools with as many as 1,200 CPUs per a single standard rack (1.2 kilo cores! — I like the sound of that!)

Of course, these remarks are more valid for the commercial data centers. In the scientific and technical segments of computing, there are already many existing superclusters using thousands of processors.

Khan explains how this trend towards large numbers of processors will influence application design and offer new possibilities for managing a data center.

Deployment and administration of applications would also change dramatically. Do you ever worry about how much storage an individual user might need? Probably not. [...] Do you ever worry about the utilization level of an individual byte of memory? I hope not. You have so many bytes that you measure utilization at the aggregate level.

Logically, he doesn’t forget to mention his company.

If you had hundreds of CPUs in a miniaturized “big-iron” system that were available to your applications, you could adopt the same strategy for applications. No need to plan capacity for each individual application. Let all of your users share a huge compute pool and plan capacity across many applications. In the process, you also fundamentally change the economics of computing. Well, that’s exactly what Azul Systems is pioneering.

This is a whole new way of looking at the CPU, and therefore, the function of “compute.” This approach is gaining mainstream acceptance. The industry has reached 2 or 4 CPUs on a chip for large symmetric multiprocessing (SMP) systems; and for systems limited to one chip, tens of functional units in one CPU. Some companies have announced future chips with as many as 8 CPUs on a single chip. With 24 CPUs on a chip that can be used in an SMP system, Azul has already set the bar much higher. And that’s just the beginning!

And before wishing us the best for 2005, Khan imagines a possible title for a 2005 press release.

Poly-core Technology to Enable Kilo Core Processing. Happy Apps Hail Freedom!!

Like in September 2004, I still don’t know if Azul Systems will be successful. But at least, I admire them for their innovative approach to computing with the additional risk of using their own custom chips.

Sources: Shahin Khan, VP and CMO at Azul Systems, for The Register, January 11, 2005; and various websites

Related stories can be found in the following categories.

• Chips
  


• Computers
  


• Hardware
  


• IT
  


• Software
  

The European Union has now 25 members — and 20 official languages, a nightmare for translators. Anticipating this, the EU started three years ago a 4-million euro project, TransType2, which is currently under test with results exceeding the original goals. In this article, the EU’s Information Society Technologies (IST) reports productivity gains in excess of 30% above traditional methods. The system mixes the advantages of both computer-assisted translation (CAT) and machine translation (MT). When you use the computer-assisted system, you start your translation, and several suggestions are offered to you while you’re typing, reducing your number of keystrokes and saving you time. Today, TransType2 allows bidirectional translations between English, French, German, and Spanish. Other European languages could easily been added. The EU is now thinking to bring this tool to us  either as a commercial product or a Web service. Read more…

Before going further, here are two screenshots illustrating the TransType2 concept (Credit: TransType2 project).

The first one comes from a page containing other screen captures while the second has been picked from this animated GIF image.

Now, let’s look at the introduction of the IST Results article.

Due to end in February, the 36-month IST programme project has drawn on two of the most commonly used translation technologies developed to date: Computer-Assisted Translation (CAT), in which human translators work in unison with a computer; and Machine Translation (MT), in which the computer handles the entire process. While both techniques have advantages and drawbacks, TransType2 has “used the best of both worlds” says project manager José Esteban at Atos Origin in Spain.

And here is how the system works.

The system works by providing translators with suggestions to complete sentences as they type which can be incorporated simply and rapidly, reducing the number of keystrokes needed to complete a translation. The suggestions are created based on statistical models of translated texts, used by the MT engines to predict the words and phrases that will come next.

The question is: does this system help to produce high quality translations faster than other methods?

Based on the work of two previous projects TransType (Canadian government-funded) and EuTrans (EU-funded), TransType2 offers significant benefits over existing techniques. Trials currently underway with two translation agencies in Canada and Spain are showing results that could be better than the project partners first expected.

“We originally thought the system would increase productivity by between 15 or 20 per cent, but in some cases we’re seeing gains in excess of 20 per cent and as high as 25 or 30 per cent,” Esteban says. “Once translators have familiarised themselves with the system the productivity increases start to become noticeable almost immediately.”

Not surprisingly, TransType2 works well with structured documents, such as technical, political or legal ones and is not very good with literary works. Still, it looks like a promising tool for the people in charge of translating the huge amounts of texts produced by the EU.

Sources: IST Results, January 12, 2005; and various other websites

Related stories can be found in the following categories.

• Computers
  
  
• Education
  
  
• Innovation
  
  
• Software