We are currently collecting signatures from ISBA members who would like to support the creation of this section.
If you are interested in learning more or would like to add your support, please reach out at kleinlab@scc.kit.edu with your full name, affiliation, and email address.
Bayesian Structural Learning (BSL) refers to methods that uncover the hidden structure of complex stochastic systems while explicitly quantifying uncertainty. This is essential in modern data science: it allows us to reason about competing structural explanations, assess their implications, and make robust decisions under uncertainty.
BSL captures both:
1. the core statistical activity of modeling and understanding structural aspects of high‑dimensional and complex data, and
2. the Bayesian principles that enable coherent uncertainty quantification, principled model comparison, and transparent integration of prior knowledge.
We believe that these ideas are currently spread across multiple ISBA Sections; but not explicitly represented by any of them; motivating the need for a dedicated BSL Section.
The new BSL Section aims to provide an inclusive home for researchers working on methods for high‑dimensional or structurally complex data, including:
* flexible regression models (latent variable & hierarchical approaches)
* dependence modeling (graphical models, copulas, networks)
* causal inference
* spatio-temporal processes
* factor models
* clustering and mixture modeling
If you support this initiative, please consider sharing this post with colleagues who might be interested.
Thank you very much for your time and support.