In a bid to master automotive autonomy, Uber is launching a new artificial-intelligence effort. Under the simple name of AI Labs, Uber’s artificial-intelligence research will initially be led by a 15-member team from Geometric Intelligence, an AI research startup. Its acquisition by Uber was announced along with the new initiative.
According to a blog post written by Jeff Holden, Uber’s chief product officer, AI Labs will attempt to solve a number of challenges related to autonomous vehicles. Geometric Intelligence, on its own website, says its goal is “to redefine the boundaries of machine learning through innovative, patent-pending techniques that learn more efficiently from less data.”
In other words, Geometric Intelligence wants to replace today’s dominant deep-machine-learning philosophy, a data-heavy approach that relies on the recognition of patterns in digital data. As Geometric Intelligence CEO and AI Labs director Gary Marcus told MIT Technology Review, “[An autonomous car] might have enough information to predict what happens at nine in the morning, but what happens at 2 a.m. and there’s less data?” The takeaway? Geometric Intelligence and AI Labs will look to create artificial intelligence that is able to learn more efficiently with less available data.
Speaking at the MIT Technology Review’s EmTech Digital conference, Marcus suggested the company already is well on its way to changing the way machines learn. In Geometric Intelligence’s early testing, he revealed that its AI was able to learn more quickly using deep learning.
Still, Geometric Intelligence’s AI will have a challenge ahead of it to navigate the incredibly dynamic world of autonomous vehicles. Nevertheless, look for Uber to report any breakthroughs at AI Labs and to apply any learnings to its autonomous-vehicle test fleet currently prowling the streets of Pittsburgh.
from Car and Driver BlogCar and Driver Blog http://ift.tt/2ghqDjd
via IFTTT
0 comments:
Post a Comment