## Abstract

Bayesian Belief Networks (BBNs) provide a way to model problems involving uncertainty. BBNs are directed acyclic graphs whose nodes are uncertain variables, and whose edges represent probabilistic dependencies between

these variables. Each node has a conditional probability function associated with it, that models the relationship between the node and its parents. BBNs are used quite widely for diagnostics, particularly medical diagnosis, where they can be used to infer the most likely illness given data about a patient and their symptoms. BBNs may also be used to solve problems in other types of system where the relationship between variables is uncertain. Many such applications will often require the use of BBNs in highly-dependable roles, where the information they provide could be critical to the safety of the system. In this paper we explore the issues associated with the use of BBNs in such critical roles, and suggest approaches to address these issues.

these variables. Each node has a conditional probability function associated with it, that models the relationship between the node and its parents. BBNs are used quite widely for diagnostics, particularly medical diagnosis, where they can be used to infer the most likely illness given data about a patient and their symptoms. BBNs may also be used to solve problems in other types of system where the relationship between variables is uncertain. Many such applications will often require the use of BBNs in highly-dependable roles, where the information they provide could be critical to the safety of the system. In this paper we explore the issues associated with the use of BBNs in such critical roles, and suggest approaches to address these issues.

Original language | English |
---|---|

Publication status | Published - 2005 |