Bayesian Structure Learning by Recursive Bootstrap

Authors: Raanan Y. Rohekar, Yaniv Gurwicz, Shami Nisimov, Guy Koren, Gal Novik

NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We empirically demonstrate that the proposed algorithm scales well to hundreds of variables, and learns better MAP models and more reliable causal relationships between variables, than other state-of-the-art-methods. (Abstract); 4 Experiments We use common networks and datasets to analyze B-RAI in three aspects: (1) computational efficiency compared to classic bootstrap, (2) model averaging, and (3) model selection. Experiments were performed using the Bayes net toolbox (Murphy, 2001).
Researcher Affiliation Industry Raanan Y. Rohekar Intel AI Lab raanan.yehezkel@intel.com Yaniv Gurwicz Intel AI Lab yaniv.gurwicz@intel.com Shami Nisimov Intel AI Lab shami.nisimov@intel.com Guy Koren Intel AI Lab guy.koren@intel.com Gal Novik Intel AI Lab gal.novik@intel.com
Pseudocode Yes Algorithm 1: Construct a graph generative tree, T; Algorithm 2: Sample a CPDAG from T
Open Source Code No No explicit statement about providing open-source code for the described methodology (B-RAI) is found. The paper only mentions using an existing toolbox.
Open Datasets Yes We use common networks2 and datasets3 to analyze B-RAI... 2www.bnlearn.com/bnrepository/ 3www.dsl-lab.org/supplements/mmhc_paper/mmhc_index.html
Dataset Splits No The paper specifies training and test dataset sizes (e.g., '500 samples for training' and '5000 samples for calculating the posterior predictive probability') but does not mention a separate validation split or explicit methodology for defining one.
Hardware Specification No No specific hardware details (e.g., CPU/GPU models, memory, or specific computing environments) used for running experiments are mentioned in the paper.
Software Dependencies No The paper mentions 'Bayes net toolbox (Murphy, 2001)' but does not specify a version number for this software or any other dependencies.
Experiment Setup Yes Conditional mutual information was used for CI testing, and BDeu with ESS = 1 for scoring. (Section 4); We apply B-RAI, with s = 3, for different sample sizes... (Section 4.1); We set γ = 1 and use the Bayesian score, BDeu (Heckerman et al., 1995). (Section 3.3.1)