Learning Performance Maximizing Ensembles with Explainability Guarantees

Authors: Vincent Pisztora, Jia Li

AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section we describe the data, model training procedures, performance evaluation metrics, and results of our experiments.
Researcher Affiliation Academia Vincent Pisztora, Jia Li Department of Statistics, Pennsylvania State University, USA
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide concrete access to source code for the methodology described.
Open Datasets Yes Following the tabular data benchmarking framework proposed by (Grinsztajn, Oyallon, and Varoquaux 2022), we conduct experiments on a set of 31 datasets (13 classification, 18 regression).
Dataset Splits Yes Each dataset is split (70%, 9%, 21%) into training, validation, and test sets respectively, following (Grinsztajn, Oyallon, and Varoquaux 2022).
Hardware Specification No Computations for this research were performed on the Pennsylvania State University s Institute for Computational and Data Sciences Roar supercomputer. This does not provide specific hardware details like GPU/CPU models.
Software Dependencies No The paper mentions types of models used (e.g., 'logistic regression', 'neural network'), but does not provide specific ancillary software details with version numbers.
Experiment Setup Yes Hyperparameter tuning for all models is done using 4-fold cross-validation, with the exception of the neural network tuning which is done using the validation set. A grid search is done to select the best hyperparameters for each model with search values available in the Appendix of the long form paper available on arxiv.org.