Building Systems that Learn inside MIT’s CSAIL
The next decade will usher in a new frontier of sophisticated systems that perform complex “human-like” tasks, with complex inferences and predictions. Using data gathered from diverse sensors and mobile devices, computing power spread across embedded devices and datacenters, as well as ubiquitous network connectivity, we will need new tools to realize the potential of learning systems. We are already seeing practical applications of these systems in areas such as autonomous vehicles and personalized health care that have the potential to transform industries and societies.
The goal of the Systems That Learn initiative at MIT MIT’s Computer Science and Artificial Intelligence Lab (MIT CSAIL)
is to accelerate the development of these systems and applications that learn. We are accomplishing this goal by combining our expertise in large scale software systems and machine learning to create new architectures, algorithms, and solutions that can understand and make sense of complex relationships unearthed by analyzing the avalanche of data available today.
In this talk we will summarize some of the guiding principles of the initiative, and discuss some of the potential risks and rewards of the rise of systems that learn. We will then talk about a few of the specific projects and solutions we are developing inside of MIT CSAIL.