shuai-huang / pe-gamp Goto Github PK

Author: Shuai Huang, The Johns Hopkins University.
Email: [email protected]

Last change: 09/03/2017
Change log: 
    v1.0 (SH) - First release (10/29/2016)
    v2.0 (SH) - Second release (09/07/2017)
    

----------------------------------------------------------------
This package contains source code for performing sparse signal recovery via PE-GAMP described in the following papers:

@INPROCEEDINGS{PEGAMP17,
author={S. Huang and T. D. Tran},
booktitle={2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
title={Sparse Signal Recovery via Generalized Approximate Message Passing with Built-in Parameter Estimation},
year={2017},
month={Sept},
}

@ARTICLE{PEGAMP_long17,
author = {{Huang}, S. and {Tran}, T.~D.},
title = "{Sparse Signal Recovery using Generalized Approximate Message Passing with Built-in Parameter Estimation}",
journal = {ArXiv e-prints},
archivePrefix = "arXiv",
eprint = {1606.00901},
year = 2016,
month = jun,
}

If you use this code and find it helpful, please cite the above paper. Thanks:)
----------------------------------------------------------------

Remember to set the parameters accordingly for your specific experiments, the code is written in MATLAB:

    1) The folder "BGM" contains the functions to perform sparse signal recovery using the proposed PE-GAMP. The input channel is assumed to be Bernoulli-Gaussian mixture (BGM) channel, the output channel is assumed to be additive white Gaussian noise (AWGN) channel.
        a) "BGM_PE_GAMP.m" is the main function used to perform sum-product PE-GAMP.
        b) "bgm_input_update.m" is the function used to estimate the input channel parameters.
        c) "awgn_output_update.m" is the function used to estimate the output channel parameters using sum-product message passing.
    
    2) The folder "BEM" corresponds to the Bernoulli-Exponential mixture (BEM) input channel and the AWGN output channel.
        a) "BEM_PE_GAMP.m" is the main function used to perform sum-product PE-GAMP.
        b) "bem_input_update.m" is the function used to estimate the input channel parameters.
        
    3) The folder "LP" corresponds to the Laplace (LP) input channel and the AWGN output channel.
        a) "SUM_LP_PE_GAMP.m" is the main function used to perform sum-product PE-GAMP.
        b) "MAX_LP_PE_GAMP.m" is the main function used to perform max-product PE-GAMP.
        c) "sum_lp_input_update.m" is the function used to estimate the input channel parameters.
        
        
    4) The folder "main" contains some functions from the GAMP MATLAB package.
    5) The folder "test_images_256" contains several image dataset.
    6) "XXX_noiseless_signal_recovery.m" contains examples to perform noiseless recovery experiments.
    7) "XXX_image_recovery.m" contains examples to perform image recovery experiments using PE-GAMP.