FoCM 2014 conference
Workshop B3 - Continuous Optimization
No date set
Compressed Sensing of Data with Known Distribution
Mateo Díaz
Universidad de los Andes, Colombia - m.diaz565@uniandes.edu.co
Compressed sensing is a technique with many important applications. For all these applications the most important parameter is the number of measurements required for perfect recovery. In this work we are able to drastically reduce the number of required measurements by incorporating information about the distribution of the data we wish to recover. Our algorithm works by minimizing an appropriately weighted $\ell^1$ norm and our main contribution is the determination of good weights.
Joint work with Mauricio Junca, Felipe Rincon, and Mauricio Velasco (Universidad de los Andes, Colombia).