Uffe Bro Kjærulff

A computational scheme for reasoning in dynamic probabilistic networks

A computational scheme for reasoning about dynamic systems using (causal) probabilistic networks is presented. The scheme is based on the framework of Lauritzen and Spiegel-halter (1988), and may be viewed as a generalization of the inference methods of classical time-series analysis in the sense that it allows description of non-linear, multivariate dynamic systems with complex conditional independence structures. Further, the scheme provides a method for efficient backward smoothing and possibilities for efficient, approximate forecasting methods. The scheme has been implemented on top of the HUGIN shell.

dHugin: a computational system for dynamic time-sliced Bayesian networks

Abstract A computational system for reasoning about dynamic time-sliced systems using Bayesian networks is presented. The system, called dHugin, may be viewed as a generalization of the inference methods of classical discrete time-series analysis in the sense that it allows description of non-linear, discrete multivariate dynamic systems with complex conditional independence structures. The paper introduces the notions of dynamic time-sliced Bayesian networks, a dynamic time window, and common operations on the time window. Inference, pertaining to the time window and time slices preceding it, are formulated in terms of the well-known message passing scheme in junction trees. Backward smoothing, for example, is performed efficiently through inter-tree message passing. Further, the system provides an efficient Monte-Carlo algorithm for forecasting; i.e. inference pertaining to time slices succeeding the time window. The system has been implemented on top of the Hugin shell.

A munin network for the median nerve-a case study on loops

Causal probabilistic networks have proved to be a useful knowledge representation tool for domains having a natural description in terms of causal relations involving uncertainty between domain concepts. This article describes a network modeling diseases affecting the median nerve. The qualitative structure of the model and the quantitative pathophysiological

Reduction of computational complexity in Bayesian networksthrough removal of weak dependences

The paper presents a method for reducing the computational complexity of Bayesian networks through identification and removal of weak dependences (removal of links from the (moralized) independence graph). The removal of a small number of links may reduce the computational complexity dramatically, since several fill-ins and moral links may be rendered superfluous by the removal. The method is described in terms of impact on the independence graph, the junction tree, and the potential functions associated with these. An empirical evaluation of the method using large real-world networks demonstrates the applicability of the method. Further, the method, which has been implemented in Hugin, complements the approximation method suggested by Jensen & Andersen (1990).

Optimal Decomposition of Probabilistic Networks by Simulated Annealing

This paper investigates the applicability of a Monte Carlo technique known as ‘simulated annealing’ to achieve optimum or sub-optimum decompositions of probabilistic networks under bounded resources. High-quality decompositions are essential for performing efficient inference in probabilistic networks. Optimum decomposition of probabilistic networks is known to be NP-hard (Wen, 1990). The paper proves that cost-function changes can be computed locally, which is essential to the efficiency of the annealing algorithm. Pragmatic control schedules which reduce the running time of the annealing algorithm are presented and evaluated. Apart from the conventional temperature parameter, these schedules involve the radius of the search space as a new control parameter. The evaluation suggests that the inclusion of this new parameter is important for the success of the annealing algorithm for the present problem.

THE HUGIN TOOL FOR PROBABILISTIC GRAPHICAL MODELS

As the framework of probabilistic graphical models becomes increasingly popular for knowledge representation and inference, the need for efficient tools for its support is increasing. The Hugin Tool is a general purpose tool for construction, maintenance, and deployment of Bayesian networks and influence diagrams. This paper surveys the key functionality of the Hugin Tool and reports on new advances of the tool. Furthermore, an empirical analysis reports on the efficiency of the Hugin Tool on common inference and learning tasks.

The Hugin Tool for Learning Bayesian Networks

In this paper, we describe the Hugin Tool as an efficient tool for knowledge discovery through construction of Bayesian networks by fusion of data and domain expert knowledge. The Hugin Tool supports structural learning, parameter estimation, and adaptation of parameters in Bayesian networks. The performance of the Hugin Tool is illustrated using real-world Bayesian networks, commonly used examples from the literature, and randomly generated Bayesian networks.

The SACSO methodology for troubleshooting complex systems

The paper describes the task of performing efficient decision-theoretic troubleshooting of electromechanical devices. In general, this task is NP-complete, but under fairly strict assumptions, a greedy approach will yield an optimal sequence of actions, as discussed in the paper. This set of assumptions is weaker than the set proposed by Heckerman et al. (1995). However, the printing system domain, which motivated the research and which is described in detail in the paper, does not meet the requirements for the greedy approach, and a heuristic method is used. The method takes value of identification of the fault into account and it also performs a partial two-step look-ahead analysis. We compare the results of the heuristic method with optimal sequences of actions, and find only minor differences between the two.

Printer troubleshooting using Bayesian networks

This paper describes a real world Bayesian network application - diagnosis of a printing system. The diagnostic problem is represented in a simple Bayes model which is sufficient under the single-fault assumption. The construction of this Bayesian network structure is described, along with guidelines for acquiring the necessary knowledge. Several extensions to the algorithms of [2] for finding the best next step are presented. The troubleshooters are executed with custom-built troubleshooting software that guides the user through a good sequence of steps. Screenshots from this software is shown.

Hybrid Propagation in Junction Trees

We introduce a methodology for performing approximate computations in complex probabilistic expert systems, when some components can be handled exactly and others require approximation or simulation. This is illustrated by means of a modified version of the familiar ‘chest-clinic’ problem.