Numerical Linear Algebra (NLA) procedures are among the primary building blocks of
machine learning algorithms. Several key concepts originating in NLA have subsequently
been adopted as first class citizens in the Machine Learning (ML) literature. Conversely,
machine learning problems are posing new mathematical and algorithmic questions for the
NLA community. This workshop will seek to foster this deep relationship further, with the
core belief that in this age of big and complex data, it is no longer optimal for NLA kernels
to be treated opaquely as a black-box in ML applications. Topics of interest include, but
are not limited to, the following:
  • Statistical Learning with Low-rank Matrix Approximations
  • Randomized Numerical Linear Algebra
  • Spectral Graph Theory and near-Linear time Linear Solvers
  • Distributed Communication-Avoiding Numerical Linear Algebra
  • Numerical Linear Algebra in Streaming and Online Settings
  • Common Big-data Software frameworks: MapReduce, MPI and Beyond
  • Applications enabled by novel interactions between ML and NLA