Bayesian Networks and their Usage within the OA-teams


  • Blixt Johan
  • Jansson Benny
  • Mossberg Karin

Publish date: 2004-01-01

Report number: FOI-R--1411--SE

Pages: 35

Written in: Swedish


The report gives an overview of what Bayesian networks (BN)are, how and when they can be used and what types of problems they are suitable for. The report can be used as a short introduction to Bayesian networks for those without previous knowledge of the method. Our emphasis has been on the applicability of the method to the types of problems that are common in the Armed Forces and that the OA-teams of Defence Analysis might be asked to solve. We have examined three different programs for building and analysing Bayesian networks, namely Hugin, Netica and GeNle, and we discuss their strengths and weaknesses. All three programs are easy to use, models can be built quickly and it is possible to build a fully functional prototype in about an hour. We believe that Bayesian networks can be used to handle several types of problems, some that are handled by the OA-teams today, like combat assessments, and others that the OA-teams do not solve today, but could possibly do in the future. Examples in the latter category are data and sensor fusion and evaluation of the value of intelligence.