[ad_1]
SAN FRANCISCO – In 2004, Geoffrey Hinton put a damper on his search for a technological idea called a neural network.
It was a way for machines to see the world around them, to recognize sounds and even to understand natural language. But scientists have spent more than 50 years working on the concept of neural networks, and the machines really could not do it.
Supported by the Canadian government, Dr. Hinton, a professor of computer science at the University of Toronto, organized a new research community with several academics who also discussed the concept. Yann LeCun, professor at the University of New York, and Yoshua Bengio, from the University of Montreal.
On Wednesday, the Association for Computing Machinery, the world's largest IT professional, announced that Drs. Hinton, LeCun and Bengio won this year's Turing Award for their work on neural networks. The Turing Award, created in 1966, is often called the Nobel Prize in Computer Science and includes a $ 1 million prize that the three scientists will share.
Over the past decade, the great idea developed by these researchers has reinvented the way technology is built, accelerating the development of facial recognition services, talking digital assistants, warehouse robots and autonomous cars. Dr. Hinton is now at Google and Dr. LeCun is working for Facebook. Dr. Bengio has signed agreements with IBM and Microsoft.
"What we have seen is nothing short of a paradigm shift in science," said Oren Etzioni, general manager of the Allen Institute for Artificial Intelligence in Seattle and leading voice in the AI. community. "The story has turned their way and I am impressed."
The neural network is a complex mathematical system that can learn discrete tasks by analyzing vast amounts of data. By analyzing thousands of old phone calls, for example, he can learn to recognize spoken words.
This allows many artificial intelligence technologies to progress at a rate hitherto impossible. Rather than manually coding behavior in systems – one logical rule at a time – computer scientists can create a technology that learns behavior largely by itself.
Dr. Hinton, 71, born in London, adopted the idea as a postgraduate student in the early 1970s, by which time most researchers in artificial intelligence were opposed. Even his own doctorate counsel questioned the choice.
"We met once a week," said Dr. Hinton in an interview. "Sometimes it ended in a screaming match, sometimes no."
Neural networks had a brief revival in the late 1980s and early 1990s. After a year of postdoctoral research with Dr. Hinton in Canada, Dr. LeCun, born in Paris, settled in the Bell Labs of AT & T, New Jersey, where he designed a network of neurons able to read letters and handwritten figures. An AT & T subsidiary sold the system to the banks and, at one point, read about 10% of all the checks in the United States.
Even though a network of neurons could read the writing and help with other tasks, it could not make much progress with the big AI tasks, such as recognizing faces and objects on pictures, identifying spoken words and understand the natural way people speak.
"They worked well only when you had a lot of training data and few areas contained a lot of training data," said Dr. LeCun, 58.
Some researchers have persisted, however, including Dr. Bengio, 55, born in Paris and who worked alongside Dr. LeCun at Bell Labs before becoming a professor at the University of Montreal.
In 2004, with funding of less than $ 400,000 from the Canadian Institute for Advanced Research, Dr. Hinton created a research program devoted to what he called "neural calculus and adaptive perception." ". He invited Dr. Bengio and Dr. LeCun to join him.
By the end of the decade, the idea had resumed its full potential. In 2010, Hinton and his students helped Microsoft, IBM and Google push the boundaries of voice recognition. Then they did the same thing with image recognition.
"He's a genius and he knows how to make an impact after another," said Li Deng, a former speech researcher at Microsoft, who brought Hinton's ideas to the company.
Dr. Hinton's discovery of image recognition was based on an algorithm developed by Dr. LeCun. In late 2013, Facebook hired the N.Y.U. teacher to build a research laboratory around the idea. Dr. Bengio resisted offers of membership to one of the big tech giants, but the research he oversaw in Montreal contributed to the growth of systems that aim to understand the natural language and technology that can generate false pictures that are not distinguishable from reality.
Although these systems have undeniably accelerated advances in artificial intelligence, they are still far removed from true intelligence. But Drs. Hinton, LeCun and Bengio believe that new ideas will come.
"We need fundamental complements to this toolbox that we have created to reach machines that work at the level of real human understanding," said Dr. Bengio.
[ad_2]
Source link