Bar codes that can be scanned using cell phones could help bring the Internet into the physical world. Already in use in parts of Asia, "physical hyperlink" technology would allow a camera-phone user to take a picture of a square bar code on an object and have their phone provide related information in the form or audio, video, or text. The required software must currently be downloaded by U.S. cell phone users. The bar codes, which can hold much more information than traditional bar codes, are being placed on some states' drivers' licenses. Although other methods for physical hyperlinking such as radio waves, computer chips, or satellite location systems have been considered, they are far more intricate and expensive than the bar codes. "The cell phone is the natural tool to combine the physical world with the digital world," says CBS executive Cyriac Roeding. A late 1990s effort to market scanning devices that could link users to information about products failed, due to a lack of consumer interest in a device that could only serve this purpose, but using cell phones for the technology seems to be a better fit. One third of U.S households with cell phones already have a camera phone, reports Forrester, and that is expected to increase in coming years. Many uses for the bar codes exist in Japan, but the most popular is for airplane tickets: A user can download the bar code to their phone and simply wave the screen over a scanner to board their flight. Users could also scan bar codes off of computer screens, if they had to go somewhere but wanted to bring the content with them. If wireless companies support the bar codes and include the reader software in their devices, they will have to choose between different bar code formats, such as Semacode, QR Code, and Qode.
Click Here to View Full Article
to the top

New Algorithms From UCSD Improve Automated Image Labeling
UCSD News (03/29/07)

Electrical engineers at the University of California, San Diego are developing an image search engine system dubbed Supervised Multiclass Labeling (SML) that can annotate and search images by analyzing their content. When evaluating an image, the system figures out the probability that various objects or "classes" are present, and creates appropriate labels. SML can also separate images by class, a process known as "image segmentations," meaning a picture of a landscape could be split up into mountains, trees, and lake sections. "Right now, Internet image search engines don't use any image content analysis," explains UCSD professor Nuno Vasconcelos. "They are highly scalable in terms of the number of images they can search but very constrained on the kinds of searches they can perform." The imaging index technique used by SML can process more images at a lower computational cost than previous methods. Not only can the system tell the difference between similar visual concepts, such as different types of bears, but it can do so for many different classes, such as bears, trees, or any concrete object. Previous systems for annotating images that have no captions have been less accurate, or have only been able to find photos similar to one they are shown or tell whether a visual concept if present in an image. During training, SML separates each image into 8-by-8 pixel squares and takes some information, known as a "localized feature," from each. All of the localized features for an image are known as a "bag of features." Researchers gather all the bags of features for a particular visual concept in a way that keeps important details of the images without keeping track of every 8-by-8 pixel square. The system has shown similarities to the way humans label images, says Vasconcelos.
Click Here to View Full Article
to the top

SC07 Tech Program Submissions Deadline Extended
HPC Wire (03/30/07)

SC07 has a new deadline of April 16, 2007, for submissions of technical papers and Gordon Bell Prize papers. As part of the two-part submission process, authors will have to register at the submission Web site ( www.sc-submissions.org) and submit an abstract by Friday, April 6, and upload manuscripts before the extended deadline. This year, technical papers will have a "blind" submission and review process. Authors should not include their names on the manuscript, and try to hide their identity. Although citations to prior work can be included in manuscripts, authors should refer to them in the third person, and also omit acknowledgements and when possible remove other identifying references. Guidelines for Gordon Bell Prize submissions can be found at sc07.supercomputing.org/html/gordonbell.html. ACM and IEEE are the sponsors for the top international conference on high performance computing, networking, and storage. SC07 is scheduled for Nov. 10-16, 2007, in Reno, Nev.
Click Here to View Full Article
to the top

Keynote: How Multicore Will Reshape Computing
EE Times (03/27/07) Goering, Richard

In his Multicore Expo keynote address, MIT professor Anant Argawal said that multicore systems-on-chip will require that designers "rethink computer architectures in a most fundamental way." He predicts that cores on a single system will number in the hundreds within the next few years. Increasing the number of processors and keeping cache sizes small will cause better results than maintaining the number of processors and making caches larger would. He said the "keep if less than linear" (KILL) rule, which means that a resource in a core should only be increased in area if the performance of the core increase proportionally, could help determine the optimal size of a cache in a specific system. For connecting cores, Argawal supported the idea of distributed meshes as opposed to busses or rings, since meshes are scalable, can be more power efficient, and offer more simple layouts. "Tiled" architecture, fully distributed with no centralized resources, is the wave of the future, while "the bus-based multicore system will fade in the next year or two," he claimed. The difficulty of multicore programming is a result of both perception, the fact that parallel programming tools are "in the dark ages," and that old programming approaches are not sufficient. ASIC designs, the ability to stream data from one computing unit to another, presents a possible solution. Core-to-core data transfer is potentially cheaper than memory access, Argawal said. Multicore devices would benefit from a "socket like" stream-based programming APIs, he said, noting that the Multicore Association's proposed Communication's API (CAPI) is this type of API. If such developments are made, the best they can do is "offer an evolutionary path," explained Argawal. "Therein lies our challenge."
Click Here to View Full Article
to the top

Linear Arrays of Nanotubes Offer Path to High-Performance Electronics
University of Illinois at Chicago (UIC) (03/26/07) Kloeppel, James E.

Researchers at the University of Illinois, Lehigh University, and Purdue University have developed a way to use dense arrays of aligned and linear carbon nanotubes as a thin-film semiconductor material that can be integrated into electronic devices. Manipulating and positioning nanotubes and achieving sufficient current outputs have been stumbling blocks for past research. These arrays could be placed on plastic or other substrates and used in technology such as flexible displays, or to improve the performance of devices built on silicon-based chips technology. "The aligned arrays represent an important step toward large-scale integrated nanotube electronics," said University of Illinois professor John A. Rogers. To build the arrays, researchers started with a wafer of single-crystal quartz on which they deposited thin strips of iron nanoparticles, which act as a catalyst for the generation of carbon nanotubes through chemical vapor deposition. As the nanotubes grow past the strips, they lock into the quartz crystal, which causes the alignment of their growth. The arrays consist of hundreds of thousand of nanotubes, which are 1 nanometer in diameter and as much as 300 microns in length. Charges move independently through the thin-film semiconductor material created by the arrays. Conventional chip-processing techniques would allow the nanotubes to be integrated into electronic devices. The researchers built a series of transistors and logic gates using the arrays and compared the arrays to individual nanotubes. "This is the first study that shows properties in scalable device configurations that approach the intrinsic properties of the tubes themselves, as inferred from single-tube studies," said Rogers.
Click Here to View Full Article
to the top

Gender Divide Growing in Computer Science
Wisconsin State Journal (03/29/07) LaRoi, Heather

Enrollment in computer science is declining just when computer science grads are becoming a valuable commodity due to an abundance of high-tech jobs and the prospects of continued job market expansion. There has been a 60 percent drop-off in the number of students who say they are interested in majoring in computer science since 2000, according to Jan Cuny with the National Science Foundation's Broadening Participation in Computing Initiative. About 10 percent of computer science bachelor's degrees at U.S. research universities are awarded to women, while the percentage of women who received bachelor's degrees in computer science at the University of Wisconsin-Madison fell from 16 percent to 9 percent between 1998 and 2006. Many experts are bewildered by the widening gender gap in computer science and IT fields, because studies indicate that efforts to get more women involved in math and science have had a generally positive effect. Cuny cites a NSF study showing that there has been a profound leakage of women in the field of computer science, as opposed to engineering and physics. One reason often given for the gender disparity is the differing reasons why men and women use computers; while men tend to see computers as devices for gaming and entertainment--and by extension, programming--women frequently view them as tools for chatting and word processing, which limits their appeal. Madison Area Technical College IT instructor Nina Milbauer says the image of IT professionals as solitary, socially maladjusted geeks is a deterring factor to girls and women that must be countered. Cuny observes that any signs of job insecurity can discourage women from entering IT, although the climate for IT jobs is more secure than ever, with the U.S. Department of Labor projecting "much faster than average" job growth for most of the chief computer science employment categories through 2014.
Click Here to View Full Article
to the top

IBM's Many Eyes Project Seeks Stats Freaks
CNet (03/29/07) Lombardi, Candace

Researchers in IBM's Visual Communication Lab plan to add more social-networking features to the Many Eyes Web site because they believe communities will be formed around the data-sharing and visualization tool. Still in public alpha, Many Eyes allows users to upload data sets from an Excel file or other common delineated files, click a few times to create a visual, and make it available to the public. Viewers can manipulate the 15 different types of data charts or visualizations to search and parse for alternative views and specific data subsets. By the summer, the IBM team plans to allow users to pull bigger versions of visualizations and tools into blogs, incorporate live updates into Web sites, add more browsing options, and create categorized forums so that people with similar interests can have a place to discuss the data. Many Eyes has grown from 20 data sets to more than 2,000, and has been used by Christian bloggers to analyze the Bible, and book enthusiasts have used it to upload data from Project Gutenberg, says Matt McKeon, a developer for the project. He believes sports statisticians will take to Many Eyes once more become aware of it, and hopes other researchers will put the Web site to good use.
Click Here to View Full Article
to the top

Binary Basics
Computerworld (03/26/07) Anthes, Gary

The binary number system is the most fundamental incarnation of computer science, and strings of zeros and ones comprise the foundation of all the software and all the data in digital computers. From the beginning, computers were designed to work directly on binary rather than decimal numbers at their lowest levels, because the binary concept has advantageous properties. British mathematician Charles Boole outlined a logic system in which the basic AND, OR, and NOT operations could form simple statements with a binary property and could be combined and stacked into the most sophisticated of logical constructs; this became the basis of computer hardware and software. In hardware, AND, OR, and NOT operations can be easily deployed as "gates," and combining enough gates generates a computer. Claude Shannon took binary a step further with his demonstration that electrical switching circuits could automatically execute Boolean logic; in essence, Shannon proved that zeroes and ones could represent all information. The science of spintronics could harbor an even more staggering breakthrough, according to University of California, Santa Barbara physics professor David Awschalom. "In contrast to zeroes and ones, with spintronics, we might go to a system with an arbitrarily large number of states," he explains. "The electron is either off or it's present with the spin pointing in one of many different directions. Each direction is a 'bit,' so that you would increase the density of information by many orders of magnitude."
Click Here to View Full Article
to the top

Tune in, Track Down
The Engineer (03/29/07)

University College London researchers are working on a radar-style detection system that uses wireless network technology to track people or objects. Today's target detection and tracking systems require targets to contain or carry an RFID or other device, can have their coverage restricted, by walls for example, and depend on the installation of expensive equipment. UCL's Wi-Fi-based tracking system could track subjects without their knowledge, be used indoors and outdoors, and use low-cost wireless hardware. "The system could be deployed anywhere with a Wi-Fi capability using the existing infrastructure," said UCL's Dr. Karl Woodbridge. "All you would have to do is to install a relatively simple receiver to build a detection system." The project is scheduled to be finished by August 2009, at which point a prototype should be ready to demonstrate detection, tracking, location, and imaging abilities using a small network. The final system would use both Wi-Fi and WiMax, and would include video surveillance to identify unusual behavior and send the information to a central unit that would evaluate the suspected threat. Signals would be received from both the transmitter and the target, but the transmitter signals would be stronger, since they have not been reflected. However, "the waveforms are designed for communications not radar detection, so they aren't necessarily ideal for passive radar transmission," Woodbridge said. "We will therefore have to develop a way to suppress the stronger signal as well as software that will be able to pinpoint the target." Wi-Fi sensors being installed for other purposes could be used for surveillance at only a slight extra cost.
Click Here to View Full Article
to the top

Jeff Hawkins, Who Invented PalmPilot, Says He's Figured Out How the Brain Works
InformationWeek (03/28/07) Wagner, Mitch

PalmPilot inventor Jeff Hawkins sees no reason why a machine cannot be made to think like a human brain. Speaking at the O'Reilly ETech Emerging Technology Conference in San Diego, Hawkins said there is no magic involved in the way human beings are able to recognize images and speech, and adapt as they learn. The neocortex is where things such as seeing and language function occur in the human brain, and although there is no differentiation in the cells in the thin layer covering this area of the brain, the various parts of the brain do different things because they are connected in different ways, he explained. The company Hawkins co-founded, Numenta, is attempting to build software that can learn and think like the human brain, and has developed a software model that is able to learn by experiencing sense data, similar to the manner in which humans do. Conference attendees saw slides that showed how the software can recognize symbols that resemble hieroglyphics. Such technology could allow car makers to build vehicles with sensors to detect dangerous driving conditions. The technology could also be used in gaming, network modeling, drug discovery, vision systems, market analysis, and business modeling.
Click Here to View Full Article
to the top

Silicon and Optics: Hybridizing for Top Performance
TechNewsWorld (03/28/07) Mello, John P. Jr.

Alcatel-Lucent Bell Labs' optical filter made from silicon can both improve the speed of fiber-optic networks and take advantage of current manufacturing processes. Today's network filters must convert an optical signal to an electrical one in order to clean it up, and then turn back into an optical signal before sending it along. What the new filter does is allow the signal to be cleaned up as an optical signal, so the high speed of optics is not hindered by electronics. "This would be a novel filter architecture even if it were made from the typical boutique photonic components," explained the labs' technical manager Sanjay Patel. "The gravy, though, is that we did this on a new silicon platform. That makes this a milestone." The union of silicon and photonics allows optics-on-a-chip systems to benefit from the silicon-based chip building infrastructure, a goal of both industry and academic researchers. The move to silicon would allow smaller, more sophisticated devices. "It opens up a whole new class of structures and architectures, which I just couldn't imagine doing in the traditional way," says Patel. Silicon photonics could increase the speed of the Internet because "you can put everything on a small chip and increase the integration level," and put many chips in a smaller, more energy-efficient device, which helps process signals much faster, says Patel. He estimates that it will be three to five years before silicon photonics are put into use in networks.
Click Here to View Full Article
to the top

Linux to Help the Library of Congress Save American History
Linux.com (03/28/07) Stutz, Michael

The preservation of American history is the goal of an ambitious project the Library of Congress is about to undertake, in which Linux-based systems will be used to digitize rare and deteriorating public domain documents and publish them online in various formats. Open source software's role in this effort will be "absolutely critical," according to the Internet Archive's Brewster Kahle. The chief element is Scribe, a book-scanning system that combines hardware and free software, and that has been wholly migrated to Linux by the Internet Archive. The documents to be scanned will be held in a Linux-based workstation at the Library of Congress and photographed by two cameras; quality assurance will be performed by a human operator, and then Scribe will transmit the images to the Internet Archive in San Francisco, where they will be processed and ultimately published online in multiple formats. Once scanned and processed, the digital versions of the documents will be made freely available online, says the Library of Congress' Dr. Jeremy E.A. Adamson. Some of the historic materials are so deteriorated and fragile that placing them in Scribe's V-shaped scanning cradle could damage them beyond repair, and Adamson says the development of a more formal classification and description of such debilitated materials and the establishment of "digitization workflows based on that classification of condition" are among the project's goals. Should new software and digitization methods be required to scan such materials, then the Library of Congress and the Internet Archive will collaborate to make the tools publicly available.
Click Here to View Full Article
to the top

GUI--Phooey!: The Case for Text Input
University of Southampton (ECS) (03/31/07) Van Kleek, Max; Bernstein, Michael; Karger, David R.

Information must be entered in order to be retrieved, but too frequently the data is not entered because it may seem too much of an effort or investment to launch and navigate through multiple applications to capture the information, or because the need to record some information can be thwarted by the lack of a natural repository. The authors offer text as a possible solution, although its unstructured form makes retrieval a sticking point; to overcome this obstacle, they have conceived of "journo," a tool designed to use lightweight text input to capture richly structured data for later retrieval and navigation in a graphical environment. The interactions the authors wanted to capture required tackling various challenges, including capturing structure from text not entered in a form, modeling the capture of a desktop state for proper tie-in to a scrap, supporting interpretation and retrieval of individual scraps of text, and blending captured data so it can be used with existing applications. Journo supports lightweight information input with a simple text input that allows users to type notes in any desirable fashion and to divide their text buffer into notes that can be freely rearranged; lowering the cost of switching applications to add notes to journo is accomplished through the provision of a number of shortcut hotkeys, and notes can be categorized through the addition of tags, which are identified syntactically as single word beginning with "@." The expression of structured information to the system in a way that is understandable to journo is offered by a simplified "pidgin" language and a lightweight triple syntax that allows the user to encompass arbitrary structural properties and relationships among entities via statements. The main technical challenges to deploying the journo design involved support for unrestricted textual input; flexibility in information structuring; integration with desktop applications; acquisition of subtext from unconstrained text; context capture and consequent selection and presentation of relevant contextual instances for facilitating refining and memory priming; and correlation among text, context, and subtext.
Click Here to View Full Article
to the top

Zeroing In
Economist (03/29/07)

As computing components reach a certain minutiae, quantum effects could have negative consequences, but researchers have found a way to use these quantum effects to benefit data storage using spintronics. The results show that a single atom could be used to store the ones and zeros that comprise binary code. A form of spintronics, known as ballistic anisotropic magnetoresistance, occurs when a magnetized wire with a width of a few atoms is placed in another magnetic field. The wire's atoms become magnetized in the direction of the field, and this direction could be used to encode a bit of data. Since electrons would be able to travel down the wire without bumping into any atoms, their spin could align itself with those of the data-storing atoms, creating a readable signal. Researchers at the University of Nebraska and the University of Strasbourg placed a narrow strip of cobalt wire on a silicon chip and made the wire into an hourglass shape at one point on its length. The "waist" of the hourglass was only one atom wide and acted as the narrow wire. Evidence showed that the signals from atoms could be read as they passed through the waist. These results are far from being ready for store shelves, but if this technology could be utilized, it could allow a superior form of compact memory.
Click Here to View Full Article
to the top

Tech Students Create Soccer-Playing Robot
Roanoke Times (VA) (03/31/07) Esposito, Greg

Virginia Tech put its robots and other innovations on display last Tuesday for its Engineering Technology Showcase. The robot that received the most attention was DARwIn (the "Dynamic Anthropomorphic Robot with Intelligence), which in its latest form was designed and programmed by a team of engineering students. A toy-sized robot that can move like a human, DARwIn was initially developed three years ago by Dennis Hong, a mechanical engineering professor and founder and director of the university's Robotics and Mechanisms Laboratory. The nearly two-foot-tall robot with human proportions makes use of sensors to keep its balance and feel with its feet, rotating and stationary cameras to locate objects, a computer and software for a brain, and motors to control its movements. DARwIn has taken second place in an international student mechanism design competition, has appeared on the cover of the robotics trade magazine Servo, and a clip of the robot on YouTube has been viewed more than 13,000 times. The robot will participate in the RoboCup robotics and artificial intelligence soccer skill competition in July at Georgia Tech.
Click Here to View Full Article
to the top

Are Your Software Programmers Coding Securely?
Computerworld (03/26/07) Vijayan, Jaikumar

A group of organizations led by the SANS Institute has launched the National Secure Programming Skills Assessment program, a series of tests designed to give companies with internal software development employees a way to test their coding skills so any flaws can be caught and corrected. Initially, four examinations will be offered, with each one testing a different type of programming language. The four areas covered are C/C++, Java/J2EE, Perl/PHP, and .Net/ASP. The exams will first be available in Washington, D.C., in August, and be made available worldwide later in the year. The necessity for a security assessment test comes from the growing need to improve programming skills while cybercriminals are becoming increasingly better at exploiting application-level vulnerabilities, many of which are the result of common coding errors such as input validation, buffer overflows, and integer errors. The program involved more than 360 organizations from the private sector, government agencies, and universities. The exams are being designed to test knowledge of basic security problems that may arise during programming, not to test advanced security knowledge. The objective is to test an individual's ability to spot coding errors and apply fundamental best practices while coding software.
Click Here to View Full Article
to the top

W3C Seeks Open Dialog About Next-Generation HTML Spec
SD Times (04/01/07)No. 171, P. 1; Worthington, David

The World Wide Web Consortium plans to recharter the HTML Working Group and include Web developers much more in the process of determining new HTML specifications. Some Web developers have expressed concern that they were not involved in the 4.01 Working Group, and ZapThink analyst Ron Schmelzer added that the heavy presence of software vendors may skew the standard to changes and other things that would benefit their products. "Involving developers in the standard will help not just the products that support the standards but also the people that have to live with them," says Schmelzer. The schedule calls for the first public working draft of the HTML revision in June, and a final version by 2010. The XHTLM 2.0 Working Group has also been rechartered, and spinning it off will allow for greater independence. Schmelzer envisions a separate community developing from users of traditional tools, best practices, and disciplines, and that it will be made up of developers who want to take a more rigorous and evolutionary approach to XHTML.
Click Here to View Full Article
to the top

To submit feedback about ACM TechNews, contact: [email protected]

To unsubscribe from the ACM TechNews Early Alert Service: Please send a separate email to [email protected] with the line

signoff technews

in the body of your message.

Please note that replying directly to this message does not automatically unsubscribe you from the TechNews list.

ACM may have a different email address on file for you, so if you're unable to "unsubscribe" yourself, please direct your request to: technews-request@ acm.org

We will remove your name from the TechNews list on your behalf.

For help with technical problems, including problems with leaving the list, please write to: [email protected]

to the top

News Abstracts © 2007 Information, Inc.