2

Black Swans & Bayesian Networks

 

Even though we may not have any actual observations, we can still speculate and hypothesize about possible rare events, i.e., we can reason on theoretical grounds as to what could possibly go wrong.

The objective of this webinar is to present Bayesian networks as a framework to merge machine-learned knowledge from data with theoretical knowledge from domain experts in order to produce a joint probability distribution that includes common and rare events at the same time.

The case study we present is addressing some of the challenges of Modern Portfolio Theory and was inspired by Rebonato & Denev's book, Portfolio Management Under Stress.

3replies Oldest first
  • Oldest first
  • Newest first
  • Active threads
  • Popular
  • Hi Stefan:

    The .xbl file does not seem to work.  Perhaps it's a difference with the latest version of Bayesialab.  Specifically I get the below errors and the system never stops even after more than an hour:

    Unable to obtain acceptable convergence!
    No exact convergence for GNK
    No exact convergence for SSW
    No exact convergence for EURN
    No exact convergence for GLOG
    No exact convergence for GNK[1]
    No exact convergence for SSW[1]
    No exact convergence for EURN[1]
    No exact convergence for GLOG[1]
     

    I've also noticed that in your version you had an option to 'Restrict Search to Selected Nodes' which makes sense to me.  However this no longer exists in the last public version of BL.  My guess is that is the source of the problem but I'm not certain.  Any ideas would be appreciated.  Thanks,

    Like
  • Hi Gabriel,

    Thanks for the feedback. It seems indeed that the option to "restrict the search to selected nodes only" is not available any more!

    As a workaround, you just have to set all the other nodes as Not Observable (Property | Cost).

    Before running the optimization, be sure you remove all the pieces of evidence that have been saved with the network. The optimization is quite time consuming on this network. Also, remember that the Genetic Algorithm is an any time algorithm that runs forever if you do not stop it manually or if you did not defined any Genetic Stop Criterion.

    Hope this helps

    Like
      • GA
      • Gabriel_Andraos
      • 2 mths ago
      • Reported - view

      Lionel Jouffe That's great.  Many thanks for the workaround.  I'll test it out.  Quick question - which one is the best optimizer for this type of problem in your opinion?

      Like
Like2 Follow
  • 2 Likes
  • 2 mths agoLast active
  • 3Replies
  • 108Views
  • 3 Following