A New System Increases the Reliability of Opinion Polls
Universidad Politecnica de Madrid (Spain) (04/26/11) Eduardo Martinez
Universidad Politecnica de Madrid researchers have developed a fuzzy neural network that uses a numerical and categorical imputation method to reconstruct incomplete data sets, which could be used to determine the voting intention of a voter that has not answered all the opinion poll questions with near 90 percent accuracy. The system, developed by Jesus Cardenosa and Pilar Rey del Castillo, also can be used for medical diagnosis or surveying using categorical variables. The system works by first defining the distances between categories using fuzzy logic. It then determines where each category is located within the different dataset spaces using the neural network. Finally, the system extends the network architecture to all the data and processes the missing data.
Tuesday, April 26, 2011
Blog: A New System Increases the Reliability of Opinion Polls
Thursday, April 21, 2011
Blog: Cyber-Security System Mimics Human Immune Response
Cyber-Security System Mimics Human Immune Response
Discovery News (04/21/11) Eric Niiler
The U.S. Department of Homeland Security's Bruce McConnell recently released a white paper that describes a healthy ecosystem of computers that work together to fight cyberthreats. McConnell says the first step to developing the ecosystem is creating a computer system that can automatically recognize and react to threats. However, a major obstacle to such a system is developing computers that can authenticate interactions, says Science Applications International Corp.'s Ross Hartman. He says researchers currently are studying new models of nature-inspired defenses as a way to protect computers from new threats. Hartman says that McConnell's paper, "Enabling Distributed Security in Cyberspace: Building a Healthy and Resilient Cyber Ecosystem with Automated Collective Action," is a positive response to rising threats and will lead to new innovations from cybersecurity experts.
Blog: The Botnets That Won't Die
The Botnets That Won't Die
Technology Review (04/21/11) Kurt Kleiner
Researchers warn that coordinated attacks on conventional botnets could lead spammers and criminal organizations to pursue more resilient communication schemes. Although conventional botnets are controlled by a few central computers, botnets that use peer-to-peer communications protocols pass messages from machine to machine. The controller inserts a command into one or more of the peers and it is spread gradually throughout the network. Some botnets using peer-to-peer communications have been implemented, but authorities have been able to infiltrate and disrupt them by spreading phony commands, files, and information. Meanwhile, Los Alamos National Laboratory's Stephen Eidenbenz and colleagues have designed and simulated a botnet that potentially would be even more difficult to shut down--one that would randomly configure itself into a hierarchy, with peers accepting commands only from machines higher up in the hierarchy, and would reconfigure the hierarchy every day.
Wednesday, April 20, 2011
Blog: iPhone Secretly Tracks User Location, Say Researchers
iPhone Secretly Tracks User Location, Say Researchers
Computerworld (04/20/11) Gregg Keizer
Apple iPhones and iPads track users' locations and store the data in an unencrypted file on the devices and on owners' computers, according to two researchers. The data is in a SQLite file on devices with 3G capability. The file, named consolidated.db, includes locations' longitude and latitude, a timestamp, and nearby Wi-Fi networks. "There can be tens of thousands of data points in this file," the researchers say. To view the location file on an iPhone remotely, an attacker would have to exploit a pair of vulnerabilities, one to hack Safari and another to gain access to the root directory, says researcher Charlie Miller. The biggest threat to users would be if the device is lost, making the data available to whoever finds it. The researchers created an application that extracts the data from a Mac and displays the location history on a map. "Why this data is stored and how Apple intends to use it--or not--are important questions that need to be explored," according to the researchers.
Tuesday, April 19, 2011
Blog: Supercomputer Cracks 'Impossible' Calculation
Supercomputer Cracks 'Impossible' Calculation
The Australian (Australia) (04/19/11) Jennifer Foreshew
Researchers at Newcastle University, IBM Australia, and the Lawrence Berkeley National Laboratory recently used the BlueGene/P supercomputer to calculate the billionth decimal digit of pi, a calculation that was thought to be unachievable. The project would have taken about 1,500 years on one central-processing unit, but it took just a few months using BlueGene/P. "What this is driving is a new attack on various classical questions about how random or how complex various bits of math are, and how best to program these things on really large environments with tens or hundreds of thousands of processors," says Newcastle professor Jon Borwein. "We may be able to put some of these algorithms together, mixing this idea of algorithmic randomness with this fairly new area called quantum randomness, using natural processes to build random things." The researchers are developing a prototype system that could lead to further advances in algorithmic randomness and quantum randomness.
Wednesday, April 13, 2011
Blog: Programming Regret for Google
Programming Regret for Google
American Friends of Tel Aviv University (04/13/11)
Tel Aviv University researchers recently launched a project aimed at developing new algorithms that will help computers minimize the distance between a desired outcome and the actual outcome, or what Tel Aviv professor Yishay Mansour calls regret. Google plans to fund the research, which is on the cutting edge of computer science and game theory. "If the servers and routing systems of the Internet could see and evaluate all the relevant variables in advance, they could more efficiently prioritize server resource requests, load documents, and route visitors to an Internet site, for instance," Mansour says. His algorithm, which is based on machine learning, minimizes the amount of virtual regret a computer might experience. "Compared to human beings, help systems can much more quickly process all the available information to estimate the future as events unfold--whether it's a bidding war on an online auction site, a sudden spike of traffic to a media Web site, or demand for an online product," Mansour says.
Wednesday, April 6, 2011
Blog: DARPA Will Spend $20 Million to Search for Crypto's Holy Grail
DARPA Will Spend $20 Million to Search for Crypto's Holy Grail
Forbes (04/06/11) Andy Greenberg
The U.S. Defense Advanced Research Projects Agency (DARPA) plans to spend $20 million over five years to find a way to both encrypt data and let it be used and manipulated. The Programming Computation on Encrypted Data (PROCEED) project would build upon the work of IBM researcher Craig Gentry, who has solved the theoretical problem of performing complex computations on encrypted data without decrypting it. Such full homomorphic encryption would enable someone to query a database without it ever knowing the content of the request. Gentry's method takes immense computational power, so DARPA wants the participating contractors and academic research teams to reduce the computing time for full homomorphic encryption by a factor of 10 million compared to its current state, or alternatively reduce it to 100,000 times the computation required for unencrypted computing. Meanwhile, Gentry says he recently discovered a less efficient version that could offer more computational shortcuts. Gentry recently received ACM's Grace Murray Hopper Award, which is awarded to the outstanding young computer professional of the year.
Tuesday, April 5, 2011
Blog: Microsoft Researchers: NoSQL Needs Standardization
Microsoft Researchers: NoSQL Needs Standardization
InfoWorld (04/05/11) Joab Jackson
The growing number of non-relational structured query language (NoSQL) databases needs standardization in order to thrive, according to Microsoft researchers. "Programming, deploying, and managing NoSQL solutions requires specialized and low-level knowledge that does not easily carry over from one vendor's product to another," write Microsoft's researchers Erik Meijer and Gavin Bierman in the April issue of Communications of the ACM. The researchers developed coSQL, a mathematical data model and standardized query language that could be used to combine NoSQL and SQL data models. The researchers say that NoSQL could benefit from the same type of standardization that SQL experienced in the early 1970s. "Just as Codd's discovery of relational algebra as a formal basis for SQL ... propelled a billion-dollar industry around SQL, we believe that our categorical data-model formalization and monadic query language will allow the same economic growth to occur for coSQL key-value stores," the researchers write.