Computing Divergences between Discrete Decomposable Models
Authors: Loong Kuan Lee, Nico Piatkowski, François Petitjean, Geoffrey I. Webb
AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Runtime Comparison with mcgo Recall that a method already exists for computing the KL divergence between 2 BNs (Moral, Cano, and G omez-Olmedo 2021) which we will refer to as mcgo. Also note that it is possible to take a distribution represented by a BN and, in exchange for some loss in independence information, represent it using a DM instead (Koller and Friedman 2009, p.p. 134). Therefore, one might ask, how does the practical runtime of JFComp compare to mcgo when computing the KL divergence between 2 BNs. To answer this question, we will replicate the experiment used by Moral, Cano, and G omez-Olmedo. |
| Researcher Affiliation | Academia | Loong Kuan Lee1, Nico Piatkowski2, Franc ois Petitjean1, and Geoffrey I. Webb1 1 Department of Data Science and AI, Monash University, Melbourne, Australia 2Fraunhofer IAIS, Schloss Birlinghoven, 53757 Sankt Augustin, Germany |
| Pseudocode | No | The paper describes algorithmic procedures, but does not include any clearly labeled pseudocode or algorithm blocks. |
| Open Source Code | Yes | The repository for the implementation for JFComp can be found at: https://lklee.dev/pub/2023-aaai/code |
| Open Datasets | Yes | They chose a set of BNs from the bnlearn (Scutari 2010) repository (https://www.bnlearn.com/bnrepository/) to sample from and estimated a second BN from these samples. |
| Dataset Splits | No | The paper mentions sampling data and learning Bayesian Networks but does not provide specific train/validation/test dataset splits, percentages, or absolute sample counts for reproducibility. |
| Hardware Specification | Yes | We run the experiments on an Intel NUC-10i7FNH with 64GB of RAM. |
| Software Dependencies | No | The paper states 'The implementation of both methods are in Python and use the pgmpy library', but does not specify version numbers for Python or the pgmpy library. |
| Experiment Setup | Yes | We repeat this 10 times in order to get an estimate of both methods runtime in seconds. |