Suvrit's Research and Academic Interests

My primary interests lie in machine learning and data mining as approached from an optimization theory perspective. I am also interested in a variety of mathematics and computer science problems. Here is a summary of some of the things that I am currently really interested in. If you think that you would be interested in some of these things, and want to talk to me about a potential collaboration --- do not hesitate to contact me at: Suvrit Sra
I. Medical Image Reconstruction
  Positron Emission Tomography (PET)
  Magnetic Resonance Imaging
II. Metric Learning
III. Optimization, in particular for Machine Learning Problems
  Large-scale Linear Programming
  Convex and non-convex quadratic programming
  Large scale SVMS.
  Semidefinite Progamming
  Second-order cone programs
  Non-convex optimization
  Large-scale classification and regression software
  Variational methods:
IV. Compressed Sensing
V. Non-negative matrix approximation
VI. Modeling data using directional distributions (von Mises-Fisher, Watson, etc.)
VII. Scientific Computing
  Numerical Linear Algebra -- dense and sparse
  Numerical and asymptotic approximations; root finding
  Special function computation, e.g., BesselI, 1F1, etc.
  Multi-precision Computation
  SAGE
VIII. Machine Learning for Systems Problems (SysML)
 

Other Academic Interests

  1. Matrix Analysis
  2. ConvexAnalysis
  3. Discrete Mathematics
  4. Probability and Statistics
  5. Computer Algebra
Suvrit Sra
Last modified: Thu Jan 31 15:07:07 CET 2008