Blog: Less Clumsy Code for the Cloud

Less Clumsy Code for the Cloud
Technology Review (12/16/09) Naone, Erica

Researchers at the University of California, Berkeley are working on a project called BOOM, which is developing new programming techniques for cloud computing. BOOM researchers hope to make cloud computing more efficient by using database programming techniques originally developed in the 1980s, which are designed to collect large data sets and process them in various ways. "We can't keep programming computers the way we are," says Berkeley professor Joseph Hellerstein. "People don't have an easy way to write programs that take advantage of the fact that they could rent 100 machines at Amazon." Bloom researchers are adapting an old language called Datalog to develop Bloom, a new language that would provide an easier way for programmers to work with cloud computing resources. The group also is creating a Bloom library that can be used with popular languages such as Java and Python. Oxford University professor Georg Gottlob, a Datalog expert, says the language may have been ahead of its time, but is gaining in popularity with the rise of distributed computing applications.

View Full Article

Blog: Researchers Make Significant Advances in Molecular Computing

Researchers Make Significant Advances in Molecular Computing
University of Kent (12/10/09)

The fundamental limits of molecular computing have been defined by researchers at the University of Kent, which published their findings in the Journal of the Royal Society Interface. The research also discusses how fast molecular computers can perform a computation, which must be addressed in order to design machines that use components of organisms to run calculations inside living cells. The metabolic rate or the ability to process energy would determine the speed of bio-molecular computers, says Kent's Dominique Chu. "One of our main findings is that a molecular computer has to balance a trade-off between the speed with which a computation is performed and the accuracy of the result," Chu says. "However, a molecular computer can increase both the speed and reliability of a computation by increasing the energy it invests in the computation." He says this energy could be derived from food sources. Moreover, he believes the findings have the potential to be of practical importance for computing in general.

View Full Article

Blog: Optimism as Artificial Intelligence Pioneers Reunite

Optimism as Artificial Intelligence Pioneers Reunite
The New York Times (12/08/09) P. D4; Markoff, John

An optimistic outlook has returned to the field of artificial intelligence (AI) 45 years after the pronouncement by computer scientist John McCarthy that a thinking machine could be created within a decade. Fueling the renewed optimism is rapid progress in AI technologies. More than 200 of the Stanford Artificial Intelligence Laboratory's (SAIL's) original scientists recently convened for a reunion, where the optimism was palpable. On hand were such luminaries as Don Knuth, who wrote the definitive texts on computer programming, and spell-checker designer Les Earnest. Other SAIL alumni included Raj Reddy and Hans Moravec, who made important foundational contributions to speech recognition and robotics at Carnegie Mellon University. The development of the graphical user interface was based on the philosophy of simplicity defined by SAIL veteran Larry Tesler, while McCarthy, who was SAIL's director, developed the LISP programming language and the time-sharing approach to computers prior to joining the laboratory. The strides that AI has made in recent years is especially apparent at Stanford, where a team of researchers developed an autonomous vehicle that successfully traversed 131 miles of mountain roads to win the 2005 Grand Challenge held by the U.S. Defense Advanced Research Projects Agency. "We are a first-class citizen right now with some of the strongest recent advances in the field," says current SAIL director and Stanford roboticist Sebastian Thrun.

View Full Article

Blog: Rethinking Artificial Intelligence

Rethinking Artificial Intelligence
MIT News (12/07/09) Chandler, David L.

The Massachusetts Institute of Technology (MIT) is embarking on the Mind Machine Project (MMP), an initiative led by artificial intelligence (AI) pioneers to create new breakthroughs by rethinking fundamental AI assumptions. "Essentially, we want to rewind to 30 years ago and revisit some ideas that had gotten frozen" while fixing basic mistakes made over the years, says MIT professor Neil Gershenfeld. He says the MMP aims to specifically address the three biggest quagmires in AI research--the modeling of thought, the reliable simulation of memory, and bridging the gap between computer science and physical science. Tackling the first challenge entails establishing what Gershenfeld calls "an ecology of models" so that problem-solving can be facilitated in multiple ways. Addressing the memory issue involves teaching computers to learn to reason while incorporating rather than excluding inconsistency and ambiguity. The third AI research area requires a new programming approach called reconfigurable asynchronous logic automata, whose goal is to "re-implement all of computer science on a base that looks like physics," representing computations "in a way that has physical units of time and space, so the description of the system aligns with the system it represents," Gershenfeld says. One of the projects the MMP group is developing is a brain co-processor, an assistive system designed to help people with cognitive disorders by monitoring a person's activities and brain functions, determining when he or she requires help, and supplying precisely the right piece of information at the right time.

View Full Article

Blog: Researchers Build Artificial Immune System to Solve Computational Problems

Researchers Build Artificial Immune System to Solve Computational Problems
PhysOrg.com (12/03/09) Zyga, Lisa

Oklahoma State University (OSU) researchers have published a study on the use of artificial immune systems (AIS) in evolutionary algorithms. By copying the way a living body acquires immunity to disease through vaccination, researchers have designed an AIS to more efficiently solve optimization problems. The results show that the biologically motivated approach is better at exploring a greater amount of space than previous methods. Unlike previous forms of AIS, the OSU system capitalizes on the way that vaccines can improve the performance of the immune system. Vaccines enable immune systems to detect new, weakened antigens and develop a biological memory so they can recognize the same antigen in the future. The researchers drew inspiration from how vaccines work in designing the new AIS. They inject the AIS with certain points in the decision space that act as a weak antigen, or vaccine. When comparing the new algorithm, called Vaccine-AIS, to other types of AIS, the researchers found that Vaccine AIS outperformed the others by locating the optimum point in a plot in fewer evaluations. "AIS was originally designed for data mining, anomaly detection, and the like," says OSU's Gary Yen. "Its use as an optimization tool is a very young research area but its performance is drawing interest from researchers."

View Full Article