Bayesian Network Engineering: Recursive Unfolding of Causal Interaction Models
Silja Renooij
Date: 16:00 – 16:30, Thursday, 24.06.2021
Location: MS Teams ICS Colloquium
Title: Bayesian Network Engineering: Recursive Unfolding of Causal Interaction Models
Abstract: A Bayesian network (BN) is an easy-going member of the family of Probabilistic Graphical Models (PGMs), in the sense that their interpretation is relatively intuitive which facilitates their application, especially in data poor domains. After a very brief introduction into PGMs and a high-level overview of my research in general, I will present our recently proposed method for engineering BN subnetworks that follow the assumptions of so-called causal interaction models. In doing so, I will specifically focus on the popular interaction model known as the (leaky) noisy-OR.