Analysis of Robust PCA via Local Incoherence

Authors: Huishuai Zhang, Yi Zhou, Yingbin Liang

NeurIPS 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we provide numerical experiments to demonstrate our theoretical results. In these experiments, we adopt an augmented Lagrange multiplier algorithm in [17] to solve the PCP. We set λ = 1/pn log n. A trial of PCP (for a given realization of error locations) is declared to be successful if ˆL recovered by PCP satisfies kˆL Lk F /k Lk F 10 3.
Researcher Affiliation Academia Huishuai Zhang Department of EECS Syracuse University Syracuse, NY 13244 hzhan23@syr.edu Yi Zhou Department of EECS Syracuse University Syracuse, NY 13244 yzhou35@syr.edu Yingbin Liang Department of EECS Syracuse University Syracuse, NY 13244 yliang06@syr.edu
Pseudocode No The paper describes the steps of the proof and construction (e.g., Z0, Zk, RΓk Zk 1) but does not provide a formally labeled pseudocode or algorithm block.
Open Source Code No The paper does not provide any statement about releasing source code or links to a code repository.
Open Datasets No The paper does not use pre-existing public datasets. It describes generating low-rank matrices using "Bernoulli model", "Gaussian model", and "Cluster model" for simulations, but does not provide access information for these generated datasets.
Dataset Splits No The paper discusses performing "50 trials of independent error corruption" and judging success if "nine trials out of ten are successful" for a given (r, ρ) pair. However, it does not specify traditional training, validation, and test dataset splits with percentages or counts for a fixed dataset.
Hardware Specification No The paper does not specify any hardware details such as CPU/GPU models, memory, or specific computing environments used for the experiments.
Software Dependencies No The paper mentions using "an augmented Lagrange multiplier algorithm in [17]" but does not specify any software dependencies with version numbers (e.g., Python, PyTorch, specific solver versions).
Experiment Setup Yes In these experiments, we adopt an augmented Lagrange multiplier algorithm in [17] to solve the PCP. We set λ = 1/pn log n. A trial of PCP (for a given realization of error locations) is declared to be successful if ˆL recovered by PCP satisfies kˆL Lk F /k Lk F 10 3. For all three low rank matrix models, we set n = 1200 and rank r = 10. For each value of , we perform 50 trials of independent error corruption and count the number of failures of PCP.