MIT researchers have developed a type of neural network that learns on the job, not just during its training phase. These flexible algorithms, dubbed « liquid » networks, change their underlying equations to continuously adapt …
January 28,2021 MIT researchers have developed a type of neural network that learns on the job, not just during its training phase. These flexible algorithms, dubbed « liquid » networks, change their underlying equations to continuously adapt to new data inputs. The advance could aid decision making based on data streams that change over time, including those involved in medical diagnosis and autonomous driving. « This is a way forward for the future of robot control, natural language processing, video processing—any form of time series data processing, » says Ramin Hasani, the study’s lead author. « The potential is really significant. » The research will be presented at February’s AAAI Conference on Artificial Intelligence. In addition to Hasani, a postdoc in the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL), MIT co-authors include Daniela Rus, CSAIL director and the Andrew and Erna Viterbi Professor of Electrical Engineering and Computer Science, and Ph.D. student Alexander Amini. Other co-authors include Mathias Lechner of the Institute of Science and Technology Austria and Radu Grosu of the Vienna University of Technology. Time series data are both ubiquitous and vital to our understanding the world, according to Hasani. « The real world is all about sequences. Even our perception—you’re not perceiving images, you’re perceiving sequences of images, » he says. « So, time series data actually create our reality. » He points to video processing, financial data, and medical diagnostic applications as examples of time series that are central to society.
