The goal for many technology companies today: determine what you, their client , then, even before asking. It is driven by something called similarity search.
"If you go to YouTube and you watch a video, they will suggest videos similar to the one you are watching. It's the search for similarity. If you go to Amazon and look for similar products to that that you are going to buy, it's the search for similarity. "
Saket Navlakha, computer scientist at the Salk Institute. He says we also do similarity searches, for example, when we scan faces in a crowd for the one we know. And even the fruit files make a version, related to the sense of smell:
"So the fly has to solve a similar problem, so to look in its database for previous experiences and previous smells that it has felt, to determine what should be the most appropriate behavioral response to that Odour."
But flies mark incoming odors differently from the way modern search algorithms analyze similarity. A small group of neurons performs a first evaluation of the odor. Then a much larger set of neurons is activated to make a final decision about the odor. Rather than the way a computer similarity search does it, taking something with a lot of dimensions, and simplifying it to a few.
So Navlakha and his colleagues twisted the computer similarity search functions to make it fly the style. And then piqued fly-inspired algorithms against conventional algorithms. And the biologically inspired code has prevailed, it is better to say "I like" of "unlike" on a similarity test.
"You know that evolution has understood, she has found a very elegant solution to this very important problem." The report is in the journal Science . [Sanjoy Dasgupta et al, A neural algorithm for a fundamental computing problem]
Navlakha says he and his team are now looking to partner with technology companies, hoping to equip machines with brain-proven problem-solving capabilities. Even though it's a fruit fly brain.
-Christopher Intagliata
[The above text is a transcript of this podcast.]
[ad_2]
Source link