Parallel Direction Method of Multipliers

Authors: Huahua Wang, Arindam Banerjee, Zhi-Quan Luo

NeurIPS 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we evaluate the performance of PDMM in solving robust principal component analysis (RPCA) and overlapping group lasso [28]. We compared PDMM with ADMM [2] or GSADMM (no theory guarantee), s ADMM [17, 26], and RBSUMM [14]. All experiments are implemented in Matlab and run sequentially. We run the experiments 10 times and report the average results.
Researcher Affiliation Academia Huahua Wang , Arindam Banerjee , Zhi-Quan Luo University of Minnesota, Twin Cities {huwang,banerjee}@cs.umn.edu, luozq@umn.edu
Pseudocode No The paper describes algorithmic steps using mathematical equations (e.g., 5-7) but does not present them within a formally labeled 'Pseudocode' or 'Algorithm' block.
Open Source Code No The paper mentions 'All experiments are implemented in Matlab' but does not provide any specific link or statement about the open-source availability of the code for the described methodology.
Open Datasets Yes A = L + S + V is generated in the same way as [17]1. ... 1http://www.stanford.edu/~boyd/papers/prox_algs/matrix_decomp.html
Dataset Splits No The paper describes the data generation process and stopping criteria for experiments, but does not provide specific details on training, validation, or test dataset splits (e.g., percentages or sample counts).
Hardware Specification No The paper states 'All experiments are implemented in Matlab and run sequentially' but provides no specific details about the hardware (e.g., CPU/GPU models, memory) used for these experiments.
Software Dependencies No The paper states 'All experiments are implemented in Matlab' but does not specify a version number for Matlab or any other ancillary software dependencies.
Experiment Setup Yes In PDMM, ρ = 1 and τi, νi are chosen according to (8), i.e., (τi, νi) = {(1/3, 1/3), (1/6, 1/6), (1/9, 1/9)} for PDMM1, PDMM2 and PDMM3 respectively. We choose the best results for GSADMM (ρ = 1) and RBSUMM (ρ = 1, α = ρ^(1/(t+10))) and s ADMM (ρ = 1). The stopping criterion is either when the residual is smaller than 10^-4 or when the number of iterations exceeds 2000.