Unique_Fac_of_Low_Rank

This deposite is the implementation of experiment results in our paper Unique Sparse Decomposition of Low Rank Matrices.

Run recover_one_vs_n_and_r.m and recover_one_vs_theta_and_r.m to reproduce left and right part of figure 1 respectively.
Run general_full_recover_L4_varing_theta.m to reproduce figure 3.
Run theta_general_full_recover_three_deflation.m and r_general_full_recover_three_deflation.m to get figure 4.

We use codes from DL_focm and LARS to compare our method with ADMM and LARS respectively in solving Sparse-PCA problem.

Codes are written by Dian Jin, Xin Bing and Yuqian Zhang. Contact dj370@scarletmail.rutgers.edu if there exist any problem.

Name		Name	Last commit message	Last commit date
Latest commit History 16 Commits
ADMM_deflation.m		ADMM_deflation.m
GD_L4.m		GD_L4.m
Gradiant_decent_L4.m		Gradiant_decent_L4.m
L4_Grad.m		L4_Grad.m
README.md		README.md
Simple_GD_L4.m		Simple_GD_L4.m
Simple_GD_L4_final.m		Simple_GD_L4_final.m
error3.m		error3.m
general_full_recover_L4_varing_theta.m		general_full_recover_L4_varing_theta.m
learn_orthobasis_adm_proj.m		learn_orthobasis_adm_proj.m
proj_orthogonal_group.m		proj_orthogonal_group.m
project1.m		project1.m
prox_L1.m		prox_L1.m
r_general_full_recover_three_deflation.m		r_general_full_recover_three_deflation.m
random_ini_X.m		random_ini_X.m
recover_one_vs_n_and_r.m		recover_one_vs_n_and_r.m
recover_one_vs_theta_and_r.m		recover_one_vs_theta_and_r.m
spca.m		spca.m
spca_error.m		spca_error.m
theta_general_full_recover_three_deflation.m		theta_general_full_recover_three_deflation.m

Jindiande/Unique_Fac_of_Low_Rank