Event-Driven Continuous Time Bayesian Networks
Authors: Debarun Bhattacharjya, Karthikeyan Shanmugam, Tian Gao, Nicholas Mattei, Kush Varshney, Dharmashankar Subramanian3259-3266
AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We demonstrate the power of the representation by applying it to model paths out of poverty for clients of City Link Center, an integrated social service provider in Cincinnati, USA. Here the ECTBN formulation captures the effect of classes/counseling sessions on an individual s life outcome areas such as education, transportation, employment and financial education.In this paper we make three major interrelated contributions: (1) we introduce a novel, interpretable yet analytically sophisticated graphical model that captures joint dynamics involving both event occurrences, modeled as a multivariate point process, and state variables, modeled as Markov processes; (2) we propose an algorithm for learning the structure and parameters of this model from data involving traces of events and transitions of state variables. We prove its consistency in the asymptotic data case whilst also demonstrating its effectiveness for limited data through experiments with synthetic data; (3) We conduct a detailed study applying our model to a real-world dataset pertaining to social service. |
| Researcher Affiliation | Collaboration | Debarun Bhattacharjya, Karthikeyan Shanmugam, Tian Gao, Nicholas Mattei, Kush R. Varshney, Dharmashankar Subramanian Research AI, IBM T. J. Watson Research Center, Yorktown Heights, NY, USA Department of Computer Science, Tulane University, New Orleans, LA, USA |
| Pseudocode | No | The paper describes the 'greedy search procedure' and 'sub-graph learning approach' in prose, but does not provide structured pseudocode or an algorithm block. |
| Open Source Code | No | The paper does not mention releasing source code for the described methodology or provide a link to a code repository. |
| Open Datasets | No | The paper states: 'The data used for this section comes from our partnership with the City Link Center in Cincinnati, Ohio, USA'. This is a specific organization's data, not explicitly stated to be publicly available with access details. For synthetic data, it notes: 'Details about the synthetic data generator are provided in Appendix B.', but does not provide public access to the generated datasets themselves. |
| Dataset Splits | Yes | Using 5-fold cross validation, we determine the optimal hyper-parameter setting by maximizing the average BIC score across folds. |
| Hardware Specification | No | The paper does not provide any specific hardware details such as GPU models, CPU types, or memory used for running the experiments. |
| Software Dependencies | No | The paper does not provide specific version numbers for any software dependencies or libraries used in the implementation or experimentation. |
| Experiment Setup | Yes | First, we configure a hyper-parameter setting for windows in Wc associated with incoming edges into X by uniformly randomly choosing a window from the list {15, 30, 60, 90, 180} days for each event label. We repeat this procedure 100 times to build various window hyperparameter configurations. |