Google Scoops up Self-Learning Neural-Network Startup for Image Searches
March 13, 2013 2:50 PM
comment(s) - last by
University of Toronto startup could help Google's search engine recognize abstract objects
Google Inc. (
) is known for scooping up promising startups, folding them into its vast internet services infrastructure. This week the company
that it was acquiring DNNresearch Inc., a one-year-old neural networks startup.
Neural networks -- artificial intelligence computing constructs that mimic animal brain neurons -- were first formulated in the 1940s, and were first investigated in the 1950s and 1960s on computing machines. But the technology fell out of favor in the early years of the digital era. It was not until the late 1980s that the technology became popular once more, thanks to the rediscovery of the backpropogation algorithm.
The latest development in neural networks over the last decade has been the creation of so-called "deep neural networks", which
utilize unsupervised learning
to optimize the execution of some desirable task. Deep neural nets can be single- or multi-layer.
DNNresearch is a deep neural network startup founded by
University of Toronto
Computer Science Department
Professor Geoffrey Hinton
and two of his graduate students, Alex Krizhevsky and Ilya Sutskever. The trio focused their unsupervised learning constructs on object recognition.
Google likely will look to put that technology to use in its image search. Google has been working with various artificial intelligence tactics to
try to recognize abstract element
s in digital images as a human brain would. It's even considered
using quantum computers
to identify objects.
DNNresearch's technology could also be applied to finding elements in audio files.
Deep neural networks can be used to recognize faces or hasty handwriting.
[Image Source: U of T]
An acquisition price was not announced. However, Google did reveal that it was giving Professor Hinton a $600,000 USD gift to support his ongoing research at U of T. The professor will now split his time between U of T, where he will continue his neural networking research, and at Google's headquarters in Mountain View, Calif., where he will work to creep the AI technology closer to the end user.
U of T President,
Professor David Naylor
, comments, "Geoffrey Hinton’s research is a magnificent example of disruptive innovation with roots in basic research. The discoveries of brilliant researchers, guided freely by their expertise, curiosity, and intuition, lead eventually to practical applications no one could have imagined, much less requisitioned. I extend my congratulations to Professor Hinton for this latest achievement."
Professor Hinton's two graduate students who cofounded the startup will be going to work full time at Google.
University of Toronto
This article is over a month old, voting and posting comments is disabled
3/13/2013 3:07:12 PM
A general method to accurately read poor quality/obscured text is AI-hard (as opposed to tuning for a specific type of mangling). While a general solution, as opposed to the current creeping incrementalism, would require replacing text captchas with something else; it'd also be a major scientific breakthrough.
3/13/2013 4:35:50 PM
There's no need to break captchas with AI. All a spammer does is set up a porn site, and mirror captchas from legit sites to people visiting the porn site before they can view the porn. An army of thousands of horny men will then solve the captchas for the spammers for free.
3/13/2013 4:52:32 PM
They already do that.
"A politician stumbles over himself... Then they pick it out. They edit it. He runs the clip, and then he makes a funny face, and the whole audience has a Pavlovian response." -- Joe Scarborough on John Stewart over Jim Cramer
Google's "Unsupervised" Self-Learning Neural Network Searches For Cat Pics
June 26, 2012, 2:30 PM
Google, D-Wave Team up to Unveil World's First Quantum Image Search
December 14, 2009, 9:40 AM
Google Releases Creative New Image, News Searches
April 21, 2009, 8:27 PM
Chris Poole Retires From Role as 4Chan After a Decade of Success, Struggles
January 23, 2015, 1:45 PM
Study Shows People are Dumb as Ever With Passwords, Still Using "123456"
January 20, 2015, 3:19 PM
Site for "Glitter as a Service" Mail Pranks, ShipYourEnemiesGlitter, Launches
January 13, 2015, 2:22 PM
OS X Yosemite Compromises Security by Retrieving Embedded Email Images
January 13, 2015, 11:30 AM
ISIS JIhadi From NZ Accidentally Shares Location on Twitter, Outs Cohorts in Selfie
January 3, 2015, 11:35 PM
Amazon's Kindle Fire HDX 8.9 Drops to $299 (30 Percent Off) for a Day
December 22, 2014, 10:57 AM
Most Popular Articles
Microsoft Shows Off Latest Windows 10 Build, Preps it for Next Week Release
January 21, 2015, 2:57 PM
Under the Hood: How DirectX 11.3 and 12 Will Supercharge Windows 10 Gaming
January 23, 2015, 12:34 PM
IDC: 2014 Sales Show PC Isn't Dead, But Desktop May be Dying
January 19, 2015, 1:50 PM
Police are Using New Handheld Radar Sensors to Peer Into Houses w/out Warrant
January 20, 2015, 1:35 PM
Report: HTC One M9 (2015) is Tied to Under Armour-Powered HTC Smartwatch
January 19, 2015, 11:10 AM
Latest Blog Posts
Sceptre Airs 27", 120 Hz. 1080p Monitor/HDTV w/ 5 ms Response Time for $220
Dec 3, 2014, 10:32 PM
Costco Gives Employees Thanksgiving Off; Wal-Mart Leads "Black Thursday" Charge
Oct 29, 2014, 9:57 PM
"Bear Selfies" Fad Could Turn Deadly, Warn Nevada Wildlife Officials
Oct 28, 2014, 12:00 PM
The Surface Mini That Was Never Released Gets "Hands On" Treatment
Sep 26, 2014, 8:22 AM
ISIS Imposes Ban on Teaching Evolution in Iraq
Sep 17, 2014, 5:22 PM
More Blog Posts
Copyright 2015 DailyTech LLC. -
Terms, Conditions & Privacy Information