Phase Retrieval Algorithm via Nonconvex Minimization Using a Smoothing Function

被引:29
|
作者
Pinilla, Samuel [1 ]
Bacca, Jorge [2 ]
Arguello, Henry [2 ]
机构
[1] Univ Ind Santander, Dept Elect & Comp Engn, Bucaramanga 680002, Colombia
[2] Univ Ind Santander, Dept Comp Sci, Bucaramanga 680002, Colombia
关键词
Non-convex problem; non-smooth function; phase retrieval; smoothing function; GRADIENT-METHOD; CRYSTALLOGRAPHY;
D O I
10.1109/TSP.2018.2855667
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
Phase retrieval is an inverse problem which consists in recovering an unknown signal from a set of absolute squared projections. Recently, gradient descent algorithms have been developed to solve this problem. However, their optimization cost functions are non-convex and non-smooth. To address the non-smoothness of the cost function, some of these methods use truncation thresholds to calculate a truncated step gradient direction. But, the truncation requires designing parameters to obtain a desired performance in the phase recovery, which drastically modifies the search direction update, increasing the sampling complexity. Therefore, this paper develops the Phase Retrieval Smoothing Conjugate Gradient method (PR-SCG) which uses a smoothing function to retrieve the signal. PR-SCG is based on the smoothing projected gradient method which is useful for non-convex optimization problems. PR-SCG uses a nonlinear conjugate gradient of the smoothing function as the search direction to accelerate the convergence. Furthermore, the incremental Stochastic Smoothing Phase Retrieval algorithm (SSPR) is developed. SSPR involves a single equation per iteration which results in a simple, scalable, and fast approach useful when the size of the signal is large. Also, it is shown that SSPR converges linearly to the true signal, up to a global unimodular constant. Additionally, the proposed methods do not require truncation parameters. Simulation results are provided to validate the efficiency of PR-SCG and SSPR compared to existing phase retrieval algorithms. It is shown that PR-SCG and SSPR are able to reduce the number of measurements and iterations to recover the phase, compared with recently developed algorithms.
引用
收藏
页码:4574 / 4584
页数:11
相关论文
共 50 条
  • [1] A Hybrid Genetic Algorithm for Nonconvex Function Minimization
    M. F. Hussain
    K. S. Al-Sultan
    Journal of Global Optimization, 1997, 11 : 313 - 324
  • [2] A hybrid genetic algorithm for nonconvex function minimization
    Hussain, MF
    AlSultan, KS
    JOURNAL OF GLOBAL OPTIMIZATION, 1997, 11 (03) : 313 - 324
  • [3] Smoothing methods for nonsmooth, nonconvex minimization
    Xiaojun Chen
    Mathematical Programming, 2012, 134 : 71 - 99
  • [4] Smoothing methods for nonsmooth, nonconvex minimization
    Chen, Xiaojun
    MATHEMATICAL PROGRAMMING, 2012, 134 (01) : 71 - 99
  • [5] Phase Retrieval Using Alternating Minimization
    Netrapalli, Praneeth
    Jain, Prateek
    Sanghavi, Sujay
    IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2015, 63 (18) : 4814 - 4826
  • [6] A Novel Iterative Algorithm for Outlier Robust Phase Retrieval via Majorization Minimization Technique
    Saini, Astha
    Arora, Aakash
    Babu, Prabhu
    IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, 2023, 72
  • [7] Phase Retrieval via Smoothing Projected Gradient Method
    Pinilla, Samuel
    Bacca, Jorge
    Angarita, Jhon
    Arguello, Henry
    2018 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2018, : 3964 - 3968
  • [8] Smoothing Nonlinear Conjugate Gradient Method for Image Restoration Using Nonsmooth Nonconvex Minimization
    Chen, Xiaojun
    Zhou, Weijun
    SIAM JOURNAL ON IMAGING SCIENCES, 2010, 3 (04): : 765 - 790
  • [9] PRIME: Phase Retrieval via Majorization-Minimization
    Qiu, Tianyu
    Babu, Prabhu
    Palomar, Daniel P.
    IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2016, 64 (19) : 5174 - 5186
  • [10] Nonconvex nonsmooth optimization via convex–nonconvex majorization–minimization
    A. Lanza
    S. Morigi
    I. Selesnick
    F. Sgallari
    Numerische Mathematik, 2017, 136 : 343 - 381