/NORST-rmc

MATLAB implementation of "Provable Subspace Tracking from Missing Data and Matrix Completion", IEEE Transactions on Signal Processing, 2019.

Primary LanguageMATLABMIT LicenseMIT

This folder contains the code accompanying the following paper -to appear, IEEE Transactions on Signal Processing (2019).

[1] "Subspace Tracking from Missing and Outlier Corrupted Data"
     Praneeth Narayanamurthy, Vahid Daneshpajooh, and Namrata Vaswani
     arXiv:1810.03051v2 [cs.LG] 30 May 2019

List of main files:

1. wrapper_NORSTmiss_fixedsubspace.m : wrapper containing the simulated data experiments for NORST-miss in fixed Subspace
2. wrapper_NORSTmiss_changingsubspace.m : wrapper containing the simulated data experiments for NORST-miss in changing Subspace
3. wrapper_NORSTmissrob_simulateddata.m : wrapper containing the simulated data experiments for NORST-miss-rob
4. wrapper_NORSTsampeff.m : wrapper containing the simulated data experiments for NORST-samp-eff
5. wrapper_VideoAnalysis.m  : wrapper for background recovery in videos.
6. NORST.m       : main function which implements the NORST-miss-robust algorithm for subspace tracking (on simulated data)
7. NORST_random.m: main function which implements the NORST-miss algorithm for subspace tracking (on simulated data)
8. NORST_video.m : main function which implements the NORST-miss-robust algorithm for background recovery (real data-video)

Folders:

YALL1 : folder containing files to implement ell-1 minimization.
PROPACK : Linear Algebra toolbox for MATLAB
data : folder containing several video data matrices for the task of background recovery

Other files:

ncrpca : code implemented Non-convex Robust PCA, NIPS 14 downloaded from authors' website and its accompaniments lansvd, lanbpro etc
cgls : fast method to implement least squares
simpleEVD : simple Eigen Value Decomposition
Calc_SubspaceError : it calculates the distance between two subspaces

For any further questions/suggestions please contact us @ vahidd/pkurpadn [at] iastate [dot] edu