## Relieving the elicitation burden of Bayesian Belief Networks

### Cached

### Download Links

Citations: | 1 - 1 self |

### BibTeX

@MISC{Wisse_relievingthe,

author = {B. W. Wisse and S. P. Van Gosliga and N. P. Van Elst and A. I. Barros},

title = {Relieving the elicitation burden of Bayesian Belief Networks},

year = {}

}

### OpenURL

### Abstract

In this paper we present a new method (EBBN) that aims at reducing the need to elicit formidable amounts of probabilities for Bayesian belief networks, by reducing the number of probabilities that need to be specified in the quantification phase. This method enables the derivation of a variable’s conditional probability table (CPT) in the general case that the states of the variable are ordered and the states of each of its parent nodes can be ordered with respect to the influence they exercise. EBBN requires only a limited amount of probability assessments from experts to determine a variable’s full CPT and uses piecewise linear interpolation. The number of probabilities to be assessed in this method is linear in the number of conditioning variables. EBBN’s performance was compared with the results achieved by applying both the normal copula vine approach from Hanea & Kurowicka (2007), and by using a simple uniform distribution. 1

### Citations

1607 |
The magical number seven, plus or minus two: some limits on our capacity for processing information
- Miller
- 1956
(Show Context)
Citation Context ...e to assess all the probabilities for large CPTs, it can also be questioned to what extent assessors can be expected to coherently provide the probabilities at the level of detail required (see e.g. (=-=Miller 1956-=-) on the limitations of human short term memory capacity). The elicitation task thus is considered a major obstacle in the use of BBNs (Druzdzel & Van der Gaag 1995, Jensen 1995). There are two ways i... |

417 | Divergence Measures Based on the Shannon Entropy
- Lin
- 1991
(Show Context)
Citation Context ...ty of an approximation to that CPT? A measure to assess the similarity between two (discrete conditional) probability distributions, with possibly different support, is the Jensen-Shannon divergence (=-=Lin 1991-=-). Based on the Kullback-Leibler divergence, this measure does not take into account the context of the CPT, the belief network. Both Henrion (1989) and Chan & Darwiche (2002) show that inference in a... |

120 | Fundamental concepts of qualitative probabilistic networks
- Wellman
- 1990
(Show Context)
Citation Context ... pa(Xc) \ Xk. We take assignment a = {xj, . . .,xu} to be an instantiation of the set of pa(Xc) = {Xj, . . . , Xu}. Formally we define Xk ∈ pa(Xc) having a positive influence on Xc, S + (Xk, Xc), as (=-=Wellman 1990-=-): for all values xc of Xc, for all pairs of distinct values xk,n > xk,o of Xk, and for all possible assignments a¬k for the set of pa(Xc) \ Xk, P(Xc > xc | xk,n, a¬k) ≥ P(Xc > xc | xk,o, a¬k). The de... |

108 | Some practical issues in constructing belief networks - Henrion - 1987 |

81 | A computational model for causal and diagnostic reasoning in inference systems - JH, Pearl - 1983 |

37 |
Vines: A new graphical model for dependent random variables
- Bedford, Cooke
(Show Context)
Citation Context ...orm solutions have only been derived for up to three conditioning variables (parent nodes). Secondly Hanea & Kurowicka (2007) provide a method for determining a CPT based on the copula vine approach (=-=Bedford & Cooke 2002-=-) that uses similar prior information: marginal distributions and adjusted (conditional) rank correlations. This method also provides a means for deriving the CPT in the general case that the variable... |

33 | How to elicit many probabilities - Gaag, Renooij, et al. - 1999 |

29 | Elicitation of probabilities for belief networks: Combining qualitative and quantitative information - DRUZDZEL, GAAG - 1995 |

28 | When do numbers really matter - Chan, Darwiche |

28 | Building probabilistic networks: “Where do the numbers come from?” guest editors’ introduction - Druzdzel, Gaag - 2000 |

23 | Causal independence for probabilistic assessment and inference using Bayesian networks - Heckerman, Breese - 1996 |

19 | Probability elicitation for belief networks: issues to consider - Renooij - 2001 |

9 | ªUsing Sensitivity Analysis for Efficient Quantification of a BeliefNetwork,º - CoupeÂ, Peek, et al. - 1999 |

9 |
Quantification experience of a DSS for mildew management in winter wheat
- Jensen
- 1995
(Show Context)
Citation Context ...il required (see e.g. (Miller 1956) on the limitations of human short term memory capacity). The elicitation task thus is considered a major obstacle in the use of BBNs (Druzdzel & Van der Gaag 1995, =-=Jensen 1995-=-). There are two ways in which the elicitation task for discrete BBNs can be relieved. The first is to make it easier for the assessors to provide the probabilistic assessments required. Van der Gaag,... |

4 | R.: Hybrid method for quantifying and analyzing bayesian belief networks. Quality and Realiability Engineering 22(6 - Hanea, Kurowicka, et al. - 2006 |

3 | Bayesian networks for enterprise risk assessment - Bonafede, Giudici - 2007 |

2 |
Hailfinder: A Bayesian system for forecasting severe weather
- Winkler
- 1996
(Show Context)
Citation Context ...c. We will then show how to obtain the CPT from these assessments and end the section with an illustrative example of the method, taken from the Hailfinder network (Abramson, Brown, Edwards, Murphy & =-=Winkler 1996-=-). 2.1 Required assessments It is assumed that the assessor has confirmed that the values of variable Xc are ordered and that the assessor can order the values of each of the variables Xk ∈ pa(Xc) suc... |

2 |
Parameter adjustment in Bayesian networks. the generalized noisy or-gate
- Díez
- 1993
(Show Context)
Citation Context ... in elicitation effort, the assumptions necessary are strong and all the variables in the noisy-OR model need to be binary, which strongly limits the applicability of the method. The Noisy-MAX model (=-=Díez 1993-=-) can be seen as the extension of the Noisy-OR to multi-valued variables. In this model the CPT is derived from ’marginal conditional’ distributions specified for each parent: for each parent the prob... |

2 | D.: Mixed non-parametric continuous and discrete bayesian belief networks. In: Advances in mathematical modeling for reliability - Hanea, Kurowicka - 2008 |

2 | Developing Complete Conditional Probability Tables from Fractional Data for Bayesian Belief Networks - Tang, McCabe |