Sorry, you need to enable JavaScript to visit this website.

facebooktwittermailshare

On the Sensitivity of Spectral Initialization for Noisy Phase Retrieval

Abstract: 

The spectral method is an important approach for signal esti- mation that is often used as an initialization to iterative methods as well as a stand-alone estimator, where the signal is estimated by the top eigenvector of certain carefully-constructed data matrix. A re- cent line of work has characterized the asymptotic behavior of such data matrices used in spectral methods, which reveals an interesting phase transition phenomenon: there exists a critical sampling thresh- old below which the estimate of the spectral method is uninforma- tive. Furthermore, optimal preprocessing functions are developed to minimize this critical sampling threshold. In particular, most of the existing work is focused on the noiseless phase retrieval prob- lem. In this paper, our goal is to examine the sensitivity of such optimal preprocessing functions in noisy phase retrieval, when there is a mismatch between the noise model used in deriving the optimal preprocessing function and the actual noise model in practice. Our results provide important insights into the choice of preprocessing functions in spectral methods.

up
0 users have voted:

Paper Details

Authors:
Vincent Monardo and Yuejie Chi
Submitted On:
10 May 2019 - 8:05am
Short Link:
Type:
Poster
Event:
Presenter's Name:
Vincent Monardo
Paper Code:
3790
Document Year:
2019
Cite

Document Files

ICASSP_specinit_Poster.pdf

(24)

Subscribe

[1] Vincent Monardo and Yuejie Chi, "On the Sensitivity of Spectral Initialization for Noisy Phase Retrieval", IEEE SigPort, 2019. [Online]. Available: http://sigport.org/4317. Accessed: Sep. 21, 2019.
@article{4317-19,
url = {http://sigport.org/4317},
author = {Vincent Monardo and Yuejie Chi },
publisher = {IEEE SigPort},
title = {On the Sensitivity of Spectral Initialization for Noisy Phase Retrieval},
year = {2019} }
TY - EJOUR
T1 - On the Sensitivity of Spectral Initialization for Noisy Phase Retrieval
AU - Vincent Monardo and Yuejie Chi
PY - 2019
PB - IEEE SigPort
UR - http://sigport.org/4317
ER -
Vincent Monardo and Yuejie Chi. (2019). On the Sensitivity of Spectral Initialization for Noisy Phase Retrieval. IEEE SigPort. http://sigport.org/4317
Vincent Monardo and Yuejie Chi, 2019. On the Sensitivity of Spectral Initialization for Noisy Phase Retrieval. Available at: http://sigport.org/4317.
Vincent Monardo and Yuejie Chi. (2019). "On the Sensitivity of Spectral Initialization for Noisy Phase Retrieval." Web.
1. Vincent Monardo and Yuejie Chi. On the Sensitivity of Spectral Initialization for Noisy Phase Retrieval [Internet]. IEEE SigPort; 2019. Available from : http://sigport.org/4317