A ‘human brain’ created using 16,000 comps

The Times of India

London: A computer that is able to "learn" like a human brain and recognize the picture of a cat, is being touted as a breakthrough in technology by Google scientists.

The computer is based on a "neural network" of 16,000 processing cores with more than a billion interconnections , each very roughly simulating a connection in a human brain.

A team from Google's cutting-edge research lab, Google X, and Stanford University , fed the system 10 million thumbnail images taken from YouTube as "training" and then tested whether it was able to recognize 20,000 objects in new images.

According to The New York Times, it performed more than twice as accurately as any previous neural network . Among the objects the system learned to recognize was a cat, one of the most regulars star of viral clips uploaded by YouTube members. "We never told it during the training, "This is a cat" , the Telegraph quoted Google fellow Jeff Dean as saying.
"It basically invented the concept of a cat," he said.Overall, the neural network achieved 15.8% accuracy . As well as cats' faces, it learned the "concepts" of human faces and bodies, by compiling a ghostly image of their general features.
The research differed from how most artificially intelligent systems are trained, in that it was given no help by human supervisors labelling features.

"The idea is that instead of having teams of researchers trying to find out how to find edges, you instead throw a ton of data at the algorithm and you let the data speak and have the software automatically learn from the data ," Andrew Ng of Stanford University added.

Now, tools to keep tab on kids’ digital lives

When her children were ready to have laptops of their own, Jill Ross bought software that would keep an eye on where they went online. One day it offered her a real surprise. She discovered that her 16-year-old daughter had set up her own video channel. Using the camera on her laptop, sometimes in her bedroom , she and a friend were recording mundane teenage banter and broadcasting it on YouTube for the whole world to see.

For Ross, who lives outside Denver , it was a window into her daughter's mind and an emblem of the strange new hurdles of modern-day parenting. She did not mention it to her daughter; she just subscribed to the channel's updates. The daughter said nothing either; she just let Mom keep watching. "It's a matter of knowing your kids," Ross said.

Parents can now use an array of tools to keep up with the digital lives of their children, raising new quandaries . Is surveillance the best way to protect children? Or should parents trust them to share if they are scared or bewildered by something online?

The answers are as varied as parents themselves. Still, the anxieties of parenting in the digital age have spawned a mini-industry , as start-ups and established companies market new tools to track where children go online, who they meet there and what they do. Because children are glued to smartphones, the technology can allow parents to track their physical whereabouts and even monitor their driving speed. If, a few years ago, the emphasis was on blocking children from going to inappropriate sites on the family computer, today's technologies promise to embed Mom and Dad — and occasionally Grandma — inside every device that children are using, and gather intelligence on them wherever they go.

A smartphone application alerts Dad if his son is texting while driving . An online service helps parents keep tabs on every chat, post and photo that floats across their children's Facebook pages. And another scans the Web in case a child decides to try a new social network that the grown-ups have not even heard of yet.

This car can drive itself through jam

LONDON: It may sound like a James Bond movie stuff, but engineers have developed an advanced car which they say can drive itself on just the press of a button, providing much-needed relaxation to the motorist stuck in heavy traffic jams.

When the traffic jam ends and the car reaches 30mph, the auto-pilot — called "Traffic Jam Assist" — hands control of the vehicle back to the refreshed driver.

The revolutionary technology , developed by engineers at US auto major Ford, is expected to be available on several of its models within five years. Prototypes are being tested at Ford's European research and advanced engineering centre in Germany and in the US, the Daily Mail reported. Experts said its widespread adoption could help speed up traffic caught in jams by up to 37% and reduce journey times by 20% by helping cars keep pace more efficiently with the flow of the traffic. The technology works using a camera and radar behind the rear-view mirror which scans the road ahead by picking out the white lines marking the lane, plus any other traffic.

Signals are then sent to the "brains" of the system in a computer central processing unit or "black box" . Once a jam is detected, the car uses a voice command to ask the driver if they want to switch to Traffic Jam Assist.
If the answer is yes, then the car assumes command — braking to stop a collision with the car in front or to slow down to meet its speed, and then accelerating to keep up with the flow of cars in front when they move off.
It will even recognize a car that "cuts in" ahead of the vehicle in front and take appropriate braking action needed.

The system is designed only for motorways, but coping with the problems of urban traffic is just a matter of time, Ford bosses said.

Pim van der Jagt, MD of Ford's Aachen operation, said: "The car will stay in the middle of the lane even when the motorway takes a curve. It will accelerate, brake and steer itself in a jam." PTI