Scalable Sparse Covariance Estimation via Self-Concordance

Authors: Anastasios Kyrillidis, Rabeeh Karimi Mahabadi, Quoc Tran Dinh, Volkan Cevher

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results on sparse covariance estimation show the merits of our algorithm, both in terms of recovery efficiency and complexity.
Researcher Affiliation Academia Anastasios Kyrillidis, Rabeeh Karimi Mahabadi, Quoc Tran Dinh and Volkan Cevher Ecole Polytechnique F ed erale de Lausanne {anastasios.kyrillidis,rabeeh.karimimahabadi,quoc.trandinh,volkan.cevher}@epfl.ch
Pseudocode Yes Algorithm 1 Inexact SCOPT for sparse cov. estimation
Open Source Code No All approaches are carefully implemented in MATLAB code with no C-coded parts. (No statement of code availability)
Open Datasets Yes This dataset contains 2833 stocks over a trading period of 1038 days, crawled from the Yahoo Finance website1. 1http://finance.yahoo.com
Dataset Splits No The paper describes generating synthetic data and using real-world stock data, but does not provide specific train/validation/test dataset splits (e.g., percentages, sample counts, or predefined split citations) needed for reproduction.
Hardware Specification No The paper does not provide specific hardware details (e.g., CPU/GPU models, memory specifications) used for running the experiments.
Software Dependencies No All approaches are carefully implemented in MATLAB code with no C-coded parts. (No specific version numbers for MATLAB or any libraries are provided)
Experiment Setup Yes In all cases, we set Imax = 500, γ = 10 10 and = 10 8. ... Without loss of generality, we fix λ = 0.5, = 0.1 for the case n = 100 and, λ = 1.5, = 0.1 for the case n = 2000. ... All algorithms under comparison are initialized with x0 = vec(diag(b )).