Institute for Mathematical Sciences

Workshop on Computational Methods in Bio-imaging Sciences

(8 - 12 January 2018)

Venue: IMS Auditorium
Organizing Committee:
  • George Barbastathis (Massachusetts Institute of Technology)
  • Ne-Te Duane Loh (National University of Singapore)
  • Paul Thomas Matsudaira (National University of Singapore)
A sparse representation of input data requires decomposing the data under a "right system" to reduce the data size by orders of magnitude in bulk without negative impact on its important information. Such a system for dimension reduction is often a frame. In the past, frames, particularly wavelet frames, have been one of the main systems used in signal and image processing for compact representation of signals, with many successful applications. In recent years, many new types and quantities of complex datasets have emerged in scientific research, e.g., high dimensional tensor data from bio-imaging sciences, large graph data from social networks, and data on manifolds from brain imaging. This emergence raises many new challenges in frame theory and sparse representation. The aim of this workshop is to bring together researchers in various areas of data sciences (e.g. mathematics, statistics, computer science, and imaging science) to share the latest developments of this diversified field and to leverage on the synergy among different research groups. The workshop will focus on the following topics:
  • Data-driven frame theory and machine learning
  • Frame theory and sparse modeling for graph data and data on manifolds
  • Computational methods for sparse approximation
  • Mathematical understanding on deep learning
  • Applications of sparse modeling in data sciences