A New Theory for Matrix Completion

Authors: Guangcan Liu, Qingshan Liu, Xiaotong Yuan

NeurIPS 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental To verify the superiorities of the nonconvex matrix completion methods over the convex program (2), we would like to experiment with randomly generated matrices. We generate a collection of m n (m = n = 100) target matrices according to the model of L0 = BC, where B Rm r0 and C Rr0 n are N(0, 1) matrices. ... For each pair of (r0, |Ω|/(mn)), we run 20 trials, resulting in 8000 simulations in total. ... Figure 2 compares the bilinear program (9) to the convex method (2).
Researcher Affiliation Academia Guangcan Liu Qingshan Liu Xiao-Tong Yuan B-DAT, School of Information & Control, Nanjing Univ Informat Sci & Technol NO 219 Ningliu Road, Nanjing, Jiangsu, China, 210044 {gcliu,qsliu,xtyuan}@nuist.edu.cn
Pseudocode No The paper does not include a section explicitly labeled "Pseudocode" or "Algorithm", nor does it present any structured code-like blocks.
Open Source Code No The paper does not provide any explicit statements about releasing source code or links to a code repository.
Open Datasets No The paper states: "We generate a collection of m n (m = n = 100) target matrices according to the model of L0 = BC, where B Rm r0 and C Rr0 n are N(0, 1) matrices." This indicates synthetic data generation rather than the use of a publicly available dataset with concrete access information.
Dataset Splits No The paper describes generating synthetic data and running trials with varying parameters (rank, observation fraction) to evaluate the method. It does not mention traditional train/validation/test splits, which are typically used with fixed datasets.
Hardware Specification No The paper does not provide any specific details about the hardware used to run the experiments, such as CPU or GPU models.
Software Dependencies No The paper does not mention any specific software dependencies or their version numbers, such as programming languages, libraries, or frameworks used for implementation.
Experiment Setup Yes We generate a collection of m n (m = n = 100) target matrices according to the model of L0 = BC, where B Rm r0 and C Rr0 n are N(0, 1) matrices. The rank of L0, i.e., r0, is configured as r0 = 1, 5, 10, , 90, 95. ... The observation fraction is set to be |Ω|/(mn) = 0.01, 0.05, , 0.9, 0.95. For each pair of (r0, |Ω|/(mn)), we run 20 trials... Here the success is in a sense that PSNR 40d B... When p = m and the identity matrix is used to initialize the dictionary A...