Learning Bayesian Networks with Bounded Tree-width via Guided Search
Authors: Siqi Nie, Cassio de Campos, Qiang Ji
AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments Bayesian Network Scores We start the experiments by comparing the BDeu scores of structures learned by the proposed method against scores from the state-of-the-art approaches. |
| Researcher Affiliation | Academia | Siqi Nie Rensselaer Polytechnic Institute Troy NY, USA Cassio P. de Campos Queen s University Belfast Belfast, UK Qiang Ji Rensselaer Polytechnic Institute Troy NY, USA |
| Pseudocode | Yes | Algorithm 1 Find the optimal k-tree given an initial (k+1)clique. and Algorithm 2 Learning a Bayesian network structure of bounded tree-width. |
| Open Source Code | No | The paper states 'The implementation of the proposed algorithm is in Matlab', but it does not provide any link or explicit statement that its source code for the described methodology is publicly available. |
| Open Datasets | Yes | We use a collection of data sets from the UCI repository (Bache and Lichman 2013) of varying dimensionalities |
| Dataset Splits | No | The paper mentions generating samples and using datasets but does not provide specific training/validation/test split percentages, sample counts, or references to predefined splits for reproduction. |
| Hardware Specification | Yes | The implementation of the proposed algorithm is in Matlab on a desktop with 16GB of memory. |
| Software Dependencies | No | The paper states 'The implementation of the proposed algorithm is in Matlab', but it does not provide any version number for Matlab or other specific software libraries/solvers used. |
| Experiment Setup | Yes | In all experiments, we maximize the Bayesian Dirichlet equivalent uniform (BDeu) score with equivalent sample size equal to one (Heckerman, Geiger, and Chickering 1995). It has been given ten minutes of running time, same that was given to all other approximate algorithms. The exact methods (MILP and GOBNILP) are given three hours of running time. The iteration limit is set to 100 with different initial cliques. |