Analysis of Iterative Regularized Least-Square Algorithm for Phaseless Reconstruction
given at esi12 (08.10.12 15:30)
This talk presents the analysis of IRLS algorithm for vector class reconstruction from measurements y=|Ax|+noise.
The descending path navigates through a special set of rank two symmetric operators. We obtain a local
convergence result iin the absence of noise. Performance is compared to Cramer-Rao lower bounds.