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 Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
A Latent Multilayer Graphical Model For Complex, Interdependent Systems
Authors: Martin Ondrus, Ivor Cribben, Yang Feng
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We rigorously evaluate our method on both simulated and multimodal neuroimaging data, demonstrating improvements over state-of-the-art approaches. All the relevant R code implementing the method in the article is available on Git Hub. ... 6 Simulated data study ... 7 Multimodal neuroimaging data study |
| Researcher Affiliation | Academia | Martin Ondrus Neuroscience and Mental Health Institute University of Alberta Ivor Cribben Alberta School of Business University of Alberta Yang Feng School of Global Public Health New York University |
| Pseudocode | Yes | Algorithm 1 Sparse + low-rank inverse covariance estimation (SLICE) with GLASSO ... Algorithm 2 Multilayer SLICE (multi SLICE) estimation |
| Open Source Code | Yes | All the relevant R code implementing the method in the article is available on Git Hub. |
| Open Datasets | Yes | We apply multi SLICE and competitor methods to a multimodal neuroimaging dataset from Wakeman and Henson (2015). In this dataset, 16 subjects are scanned during the presentation of three different facial stimuli. ... Although we did not collect any human data as part of our neuroimaging experiments, we did use an open source data set which includes human subjects. |
| Dataset Splits | No | The paper describes simulation studies and analyzes a multimodal neuroimaging dataset from Wakeman and Henson (2015). For the neuroimaging data, it states '16 subjects are scanned during the presentation of three different facial stimuli.' and methods are applied 'independently for each subject and stimulus'. While analysis is done per subject and stimulus, no explicit training, validation, or test dataset splits are provided in the main text for either the simulated or real data. |
| Hardware Specification | Yes | Pre-processing steps and further details are in the Supplementary Materials, and all experiments are run on a M1 Mac Book Pro with 16GB of RAM using R 4.4.3. |
| Software Dependencies | Yes | all experiments are run on a M1 Mac Book Pro with 16GB of RAM using R 4.4.3. |
| Experiment Setup | Yes | To select ρ and r, we suggest a k-fold cross-validation over a grid, where the combined values are based on log-likelihood. Simulation experiments elucidating the sensitivity of multi SLICE to these choices are provided in the Supplementary Materials. |