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].
Community-Based Group Graphical Lasso
Authors: Eugen Pircalabelu, Gerda Claeskens
JMLR 2020 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The method s empirical performance is illustrated in an f MRI context, as well as with simulated examples. Keywords: community detection; graphical model; group penalty; joint graphical lasso |
| Researcher Affiliation | Academia | Eugen Pircalabelu EMAIL 1UCLouvain, Institute of Statistics, Biostatistics and Actuarial Sciences, Voie du Roman Pays 20, 1348 Louvain-la-Neuve, 2KU Leuven, ORSTAT and Leuven Statistics Research Center, Naamsestraat 69, 3000 Leuven, Belgium Gerda Claeskens EMAIL KU Leuven, ORSTAT and Leuven Statistics Research Center, Naamsestraat 69, 3000 Leuven, Belgium |
| Pseudocode | Yes | Figure 1 presents a schematic version of the steps in the proposed ADMM algorithm... We now present a step-by-step description of the algorithm. Step 1: Initialize Θ = eΘ = f X = e U 1 = e U 2 = A = I each having the dimension of S and where I is the identity matrix. |
| Open Source Code | No | The paper does not contain a direct link to a code repository or an explicit statement within its main text about releasing the source code. It mentions 'Attribution requirements are provided at http://jmlr.org/papers/v21/19-181.html' which might lead to code, but it is not a direct statement of code release within the paper. |
| Open Datasets | Yes | The proposed method is illustrated with a resting state (that is, subjects were not performing any tasks) functional magnetic resonance imaging (rsf MRI) example... The data correspond to a subset of the original data analyzed in Schmittmann et al. (2015) and have been kindly provided to us by one of the authors. |
| Dataset Splits | No | For the real-world fMRI data, the paper states, 'We analyze here the data for one subject, for which the brain activity for p = 114 regions of interest (ROIs) has been measured n = 240 times.' It describes the use of this data for application but does not specify any training, validation, or test splits. For simulated data, it describes how data was generated, not how an existing dataset was split. |
| Hardware Specification | No | The paper states, 'It took Com GGL1 for the rsf MRI example, on a standard laptop, 5.5 seconds (283 iterations) to converge when the tolerance threshold was 10 4'. 'Standard laptop' is too vague to be considered a specific hardware detail. |
| Software Dependencies | No | The paper refers to various methods and procedures (e.g., 'Graphical Lasso', 'sbm SDP', 'GSBM', 'CORD', 'cluster graphical lasso') and implies implementation but does not specify any software libraries, frameworks, or their version numbers used for the implementation. |
| Experiment Setup | Yes | Regularization parameters λn1 and λn2 have been selected using 3-fold cross-validation on a grid of 10 10 values, while to give equal importance to the contribution of S and X in the objective function, we have set λn3 = 1. For all tested competitors but CORD, one can fix the number of desired communities and the purpose of the analysis in Table 4 is to perform a sensitivity analysis when the number of communities is varied between 4 and 12. The regularization levels were (λn1, λn2, λn3) = (.4, .2, 1) for Com GGL1 and Com GGL2, λn1 = 0.395 for the two-step procedures that use the output of the graphical lasso. |