Fruit fly brains inform search engines of the future
Every day, websites and apps crunch huge sets of information to find data points that resemble each other, such as products that are similar to past purchases. These tasks are called similarity searches, and performing them well—and fast—has been an ongoing challenge for computer scientists. Salk faculty Saket Navlakha and Charles Stevens, together with collaborator Sanjoy Dasgupta of UC San Diego, have shown that fruit fly brains, in their efforts to identify similar odors, may have found a better way to perform similarity searches. Current computer algorithms find similar items by reducing the amount of information associated with each item. Fly brains do the opposite, expanding the amount of information associated with each item, which allows them to better distinguish similar from dissimilar. When the researchers applied this approach of expanding rather than reducing information associated with data to three standard datasets, they found the fly method dramatically improved search performance. They reported their findings in Science on November 9, 2017.
Featured Stories
Taking on the Big FiveCancer is not like other diseases. Most conditions have external causes—bacteria, viruses, injury—but cancer comes from inside us. Cells go rogue, divide recklessly, invade other tissues and spread throughout the body. They do things normal cells cannot do.
Dan Lewis – Intense ConnectionFew trustees have had a connection as intensely personal as new Board Chair Dan Lewis, who knows firsthand that cures, indeed, begin with Salk. Thanks to the research of Salk Professor Tony Hunter, the drug Gleevec was born. And thanks to Gleevec, Lewis survived leukemia.
Jared SmithNeuroscientist and self-described history geek Jared Smith wants to boldly go where no one has gone before. “If this were the 1400s,” asks Smith, “and we were Europeans exploring the world, where is the new world?” For Smith and colleagues in Xin Jin’s lab, the answer is simple: the brain.