Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].

Multi-Response Linear Discriminant Analysis in High Dimensions

Authors: Kai Deng, Xin Zhang, Aaron J. Molstad

JMLR 2024 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we validate the effectiveness of our approach through simulation studies and applications to benchmark datasets. ... Section 8 and Section 9, respectively.
Researcher Affiliation Academia Kai Deng EMAIL Xin Zhang EMAIL Department of Statistics Florida State University Tallahassee, Florida, USA Aaron J. Molstad EMAIL School of Statistics University of Minnesota Minneapolis, Minnesota, USA
Pseudocode Yes Algorithm 1 Blockwise coordinate descent algorithm for joint classification (16). ... Algorithm A.1 ADMM algorithm update for conditional classification (13).
Open Source Code No The paper does not provide explicit links to source code repositories or statements about the release of source code.
Open Datasets Yes Simulation studies and real data examples are presented in Section 8 and Section 9, respectively. ... The descriptions of the dataset, which were obtained from https://www.uco.es/kdis/mllresources/, are as follows. Yeast. ... Image. ... Virus GO. ... Gpositive GO. ... Gnegative GO.
Dataset Splits Yes For 100 independent replications, classification performance is evaluated on a testing data of size 1000. All tuning parameters are selected by 5-fold cross-validation. ... For each dataset, we randomly split the data 100 times and keep 20% of the samples as a testing set. ... We use 4/5 of the samples for training and the remaining 1/5 for testing.
Hardware Specification No The paper does not mention any specific hardware used for running experiments.
Software Dependencies No The above competitors are implemented by R packages msda, penalized LDA and glmnet. (No version numbers provided for R or the packages)
Experiment Setup Yes For all the simulated models, we set number of predictors p = 1000, sample size n = 30K where K = QM m=1 cm, and set the covariance Σ to be AR(0.3), unless specified otherwise. ... All tuning parameters are selected by 5-fold cross-validation.