Contextual Stochastic Block Model: Sharp Thresholds and Contiguity

Authors: Chen Lu, Subhabrata Sen

JMLR 2023 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we supplement our analytic results with some numerical experiments. In particular, we explore the validity of our results in regimes beyond ones studied so far in this paper. Finite sample recovery performance using Belief Propagation... Here we study the finite sample performance of this algorithm in depth (see Fig 2 for details).
Researcher Affiliation Academia Chen Lu EMAIL Department of Mathematics Massachusetts Institute of Technology Cambridge MA 02139, U.S.A. Subhabrata Sen EMAIL Department of Statistics Harvard University Cambridge, MA 02138, U.S.A.
Pseudocode No The paper describes algorithms such as the self-avoiding walk based estimator and a linearized Belief Propagation algorithm, but it does not present any of these in pseudocode blocks or clearly labeled algorithm sections.
Open Source Code No The paper does not contain any explicit statement about providing source code for the methodology described, nor does it provide a link to a code repository. It mentions using a 'linearized Belief Propagation algorithm proposed originally in (Deshpande et al., 2018, Section 6)' but does not provide its own implementation code.
Open Datasets No The paper does not use pre-existing public datasets. It describes its own generative model for the observed graph and covariates and then simulates data for its numerical experiments: "We set n = 800, p = 1000. The node covariates are drawn iid from an equicorrelated gaussian model..."
Dataset Splits No The paper uses simulated data based on a generative model and does not mention any explicit training, validation, or test dataset splits. The numerical experiments vary parameters of the simulation rather than using predefined data splits.
Hardware Specification No The paper does not provide any specific details about the hardware used to run the numerical experiments, such as CPU or GPU models, or memory specifications.
Software Dependencies No The paper mentions 'Belief propagation' and 'spectral algorithms' but does not specify any software names with version numbers for their implementation, nor any programming languages or libraries used.
Experiment Setup Yes We fix γ = 4/5, and vary p from 50 to 1000. Each line corresponds to the choice of µ = 3γ^2 + h, where h = 0.06, 0.04, 0.02, 0, 0.02, 0.04, 0.06, going from bottom to top... We set n = 800, p = 1000. The node covariates are drawn iid from an equicorrelated gaussian model with correlation parameter ρ.