In October 2006, Netflix offered $1,000,000 to the first team to improve upon the accuracy of its recommender tool by 10%. The contest couldn't have been scripted better. Closely watched by blogs and the popular press, the contest had a cast of thousands, drama, nail-biting finishes not once but twice, some surprising turns, and good science to boot. After three years of model-building, parameter-tweaking, and blending, AT&T researchers led the winning team.
The dust is still settling and there will be papers and discussions in the months and years to come, but some lessons for statistics, machine learning, and data mining are already apparent.
With healthcare costs skyrocketing, an aging population, and no political consensus for reform, technology may offer the fundamental healthcare change that's needed. A new generation of sensors—lightweight, inexpensive, capable of two-way communication—promises a wealth of high-quality, medical data at low cost.
But the real value comes in sharing health data with those who can interpret it and act on it. AT&T with its vast network reach can directly connect patients to doctors by securely transmitting medical data in real time. AT&T researchers are working with device-makers and standards boards to move health data seamlessly from sensors to AT&T’s broadband network.
Today’s electric grid is dated. Like a scene from a 50s family show, utility workers go house to house to read dials on electric meters. Consumers call the utility to report power outages. Dispatchers send out trucks to spot-check for line breaks. But this will all change.
Communications along with sensors and controls are about to modernize the electric grid.
AT&T is playing multiple roles, from providing the communications backbone for carrying sensor data, to enabling two-way communication between consumers and utilities. AT&T Research is working with standards organizations to make it all work, and anticipating the complexity of integrating the smart grid into the home.
In The News