Close Menu

Spatiotemporal Data Diagram

Our second research focus area concerns the study of large-scale problems where the underlying data have important and non-trivial spatial and/or temporal variations. Of specific concern are the following:

  1. Inverse-type problems where a physical sensing model relates the data available for processing to the spatio-temporal information of interest.
  2. Semi-supervised learning from time series comprised of data from multiple sources, sampled at irregular intervals, and where ground truth labels are exceedingly rare.
  3. Inferring causality from space-time data focusing on settings that match a potential-outcomes framework where the question is whether some binary treatment leads to improved outcomes of interest.
  4. Communications beyond the 5G paradigm where technologies such as massive distributed arrays and large intelligent surfaces bring about the possibility to communicate and coordinate a large amount of data in real time or under stringent latency constraints with high reliability using advanced machine learning ideas for solving problems in estimation, tracking, and optimization.

Spatiotemporal Data Diagram

Programming for this research focus will begin in the fall of 2020 with the T-TRIPODS Virtual Workshop on Spatio-Temporal Data looking at problems at the intersection of machine learning/data science and physics-based models and problems.   The event will take the form of a series of interactive seminars each comprising a 40-45 minute technical presentation followed by 20-30 minutes of discussion moderated by a member of the Tufts faculty and an associated graduate student.  The four speakers are:

  • Rachel Ward, W. A. "Tex" Moncrief Distinguished Professor in Computational Engineering and Sciences—Data Science, Associate Professor of Mathematics, University of Texas at Austin
  • Evrim Acar Ataman, Head of Department, Chief Research Scientist, Simula Research Laboratory
  • Vipin Kumar, Regents Professor and William Norris Chair in Large Scale Computing, Department of Computer Science and Engineering, University of Minnesota,
  • Andrew Stuart, Bren Professor of Computing and Mathematical Sciences, Caltech

Stay tuned for scheduling details!

For more information about this research area, please contacts any of the Tufts faculty working on this aspect of T-TRIPODS: