Description
Effective decision making for Emergency Response preparedness generally involves a large number of uncertain factors, which may be addressed through Bayesian Network modelling. Bayesian networks (BNs) provide a powerful framework for modeling uncertainty, making probabilistic inferences, and supporting decision-making in complex domains. This tutorial introduces participants to the foundational theory of Bayesian networks and guides them through a structured approach to problem representation and decision support. The session will begin with an overview of the theoretical underpinnings of Bayesian networks, including their graphical structure, conditional dependencies, and methods for probabilistic inference. Participants will engage in hands-on exercises designed to translate unstructured problems into structured Bayesian networks, allowing them to visualise relationships between variables and reason about uncertainty in a systematic way. A key challenge in building Bayesian networks is quantifying them effectively. This tutorial will cover practical approaches for eliciting expert subjective probabilities, providing a framework for integrating expert knowledge into Bayesian models. Additionally, we will explore methods for using empirical data to populate Bayesian networks, highlighting techniques for parameter learning and model refinement. Beyond technical modeling, Bayesian networks are particularly useful in multi-stakeholder decision-making contexts. This tutorial will demonstrate how BNs can facilitate structured discussions, incorporate diverse perspectives, and support negotiation by explicitly representing different stakeholders' incentives and alternatives. To ground the tutorial in real-world applications, we will examine case studies from Arctic Search and Rescue operations and climate adaptation in aquaculture. These examples will illustrate the practical value of Bayesian networks in high-stakes decision environments, where uncertainty and competing priorities must be managed effectively. By the end of this tutorial, participants will have a solid foundation in Bayesian network theory, practical experience in building and quantifying models, and insights into applying BNs for collaborative decision making. No prior experience with Bayesian networks is required, though familiarity with probabilistic reasoning will be beneficial.| Period | 18 May 2025 |
|---|---|
| Event title | ISCRAM 2025: Managing and Responding to Coastal Disasters & Climate Change |
| Event type | Conference |
| Location | Halifax, Canada, Nova ScotiaShow on map |
| Degree of Recognition | International |