In computer science, the buzzword of the day is “big data.” The proliferation of cheap, Internet-connected sensors — such as the GPS receivers, accelerometers and cameras in smartphones — has meant an explosion of information whose potential uses have barely begun to be explored. In large part, that’s because processing all that data can be prohibitively time-consuming.
Most computer scientists try to make better sense of big data by developing ever-more-efficient algorithms. But in a paper presented this month at the Association for Computing Machinery’s International Conference on Advances in Geographic Information Systems, MIT researchers take the opposite approach, describing a novel way to represent data so that it takes up much less space in memory but can still be processed in conventional ways. While promising significant computational speedups, the approach could be more generally applicable than other big-data techniques, since it can work with existing algorithms.
EECS graduate student Cynthia Sung is the paper’s third author, along with postdoc Dan Feldman and Daniela Rus, the director of MIT’s Computer Science and Artificial Intelligence Laboratory. Read the rest of the article on MITnews.