MIT and SenseTime announce effort to advance artificial intelligence research
Alliance will be part of new MIT Intelligence Quest.
MIT and SenseTime today announced that SenseTime, a leading artificial intelligence (AI) company, is joining MIT's efforts to define the next frontier of human and machine intelligence.
SenseTime was founded by MIT alumnus Xiao’ou Tang PhD ’96 and specializes in computer vision and deep learning technologies. The MIT-SenseTime Alliance on Artificial Intelligence aims to open up new avenues of discovery across MIT in areas such as computer vision, human-intelligence-inspired algorithms, medical imaging, and robotics; drive technological breakthroughs in AI that have the potential to confront some of the world’s greatest challenges; and empower MIT faculty and students to pursue interdisciplinary projects at the vanguard of intelligence research.
Neural networks everywhere
New chip reduces neural networks’ power consumption by up to 95 percent, making them practical for battery-powered devices.
Most recent advances in artificial-intelligence systems such as speech- or face-recognition programs have come courtesy of neural networks, densely interconnected meshes of simple information processors that learn to perform tasks by analyzing huge sets of training data.
But neural nets are large, and their computations are energy intensive, so they’re not very practical for handheld devices. Most smartphone apps that rely on neural nets simply upload data to internet servers, which process it and send the results back to the phone.
Now, MIT researchers have developed a special-purpose chip that increases the speed of neural-network computations by three to seven times over its predecessors, while reducing power consumption 94 to 95 percent. That could make it practical to run neural networks locally on smartphones or even to embed them in household appliances.
Energy-efficient encryption for the internet of things
Special-purpose chip reduces power consumption of public-key encryption by 99.75 percent, increases speed 500-fold.
Most sensitive web transactions are protected by public-key cryptography, a type of encryption that lets computers share information securely without first agreeing on a secret encryption key.
Public-key encryption protocols are complicated, and in computer networks, they’re executed by software. But that won’t work in the internet of things, an envisioned network that would connect many different sensors — embedded in vehicles, appliances, civil structures, manufacturing equipment, and even livestock tags — to online servers. Embedded sensors that need to maximize battery life can’t afford the energy and memory space that software execution of encryption protocols would require.
MIT researchers have built a new chip, hardwired to perform public-key encryption, that consumes only 1/400 as much power as software execution of the same protocols would. It also uses about 1/10 as much memory and executes 500 times faster. The researchers describe the chip in a paper they’re presenting this week at the International Solid-State Circuits Conference.