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 Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Learning Continuous Time Bayesian Networks in Non-stationary Domains
Authors: Simone Villa, Fabio Stella
JAIR 2016 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | A set of numerical experiments on synthetic data is used to compare the effectiveness of non-stationary continuous time Bayesian networks to that of non-stationary dynamic Bayesian networks. Furthermore, the performance achieved by non-stationary continuous time Bayesian networks is compared to that achieved by state-of-the-art algorithms on four real-world datasets, namely drosophila, saccharomyces cerevisiae, songbird and macroeconomics. |
| Researcher Affiliation | Academia | Simone Villa EMAIL Fabio Stella EMAIL Department of Informatics, Systems and Communication University of Milano-Bicocca Viale Sarca 336, 20126 Milan, Italy |
| Pseudocode | Yes | Algorithm 1 Learn KTTX Algorithm 2 Learn KNEX Algorithm 3 Tentative Allocation Algorithm 4 Learn UNEX Algorithm 5 Split Merge |
| Open Source Code | No | We acknowledge the precious help of Alex Hartemink who let us use the nsdbn jar executable program for learning ns DBN models. Furthermore, he also provided the drosophila and songbird datasets. |
| Open Datasets | Yes | A set of numerical experiments on synthetic data is used to compare the effectiveness of non-stationary continuous time Bayesian networks to that of non-stationary dynamic Bayesian networks. Furthermore, the performance achieved by non-stationary continuous time Bayesian networks is compared to that achieved by state-of-the-art algorithms on four real-world datasets, namely drosophila, saccharomyces cerevisiae, songbird and macroeconomics. The saccharomyces cerevisiae dataset is obtained from a synthetic regulatory network with 5 genes in saccharomyces cerevisiae (Cantone, Marucci, Iorio, Ricci, Belcastro, Bansal, Santini, di Bernardo, di Bernardo, & Cosma, 2009). |
| Dataset Splits | No | The paper does not provide specific train/test/validation dataset splits. For synthetic datasets, it mentions |
| Hardware Specification | No | The paper does not explicitly describe the specific hardware used to run its experiments (e.g., GPU/CPU models, memory amounts). |
| Software Dependencies | No | The paper mentions using the "nsdbn jar executable" which is a third-party tool, but it does not specify any software components or libraries with version numbers that were used in the authors' own implementation. |
| Experiment Setup | Yes | ns CTBN were learned by using the following parameters setting: Iters = 1,000, CT0 = 1,000, ζ = 0.8, z = 3, σ = 1, sp = 0.3, mp = 0.3, α = 1 and τ = 0.1 using the BDeu metric. Furthermore, for ns DBN and ns CTBN we set the maximum number of parents to 4. Structural learning experiments were performed with λc = {1, 2, 4} and λe = {5, 10, 15} for ns CTBN and λs = {1, 2, 4} and with λm = {10, 50, 100} for ns DBN. The network inference task was performed by learning ns CTBN under the UNE setting with the following parameter values λc = {0.2, 0.4, 1, 2} and λe = {0.5, 1, 2, 5}. Furthermore, we set the maximum number of parents to 2, the number of iterations to 1,000 and the number of runs to 100. |